(Topics in English Linguistics 22) Carl Bache, Niels Davidsen-Nielsen-Mastering English - An Advanced Grammar For Non-Native and Native Speakers-De Gruyter Mouton (1997) PDF

Mastering English
WDE
G
Topics in English Linguistics
22
Editor
Herman Wekker
Mouton de Gruyter
Berlin · New York
Mastering English
An Advanced Grammar for
Non-native and Native Speakers
Carl Bache
Niels Davidsen-Nielsen
Mouton de Gruyter
Berlin · New York 1997
Mouton de Gruyter (formerly Mouton, The Hague)
is a Division of Walter de Gruyter & Co., Berlin.
© Printed on acid-free paper which falls within the guidelines of the

ANSI to ensure permanence and durability.
Library of Congress Calaloging-in-Publication Data
Bache, Carl, 1953-

Mastering English : an advanced grammar for non-native
and native speakers / Carl Bache ; Niels Davidsen-Nielsen.
p. cm. - (Topics in English linguistics ; 22)
Includes bibliographical references and index.
ISBN 3-11-015535-4 (alk. paper)
ISBN 3-11-015536-2 (pbk)
I.English language - Grammar. I. Davidsen-Nielsen,
Niels. II. Title. III. Series.
PE1106.B27 1997
428.2'4-dc21 97-23493
CIP
Die Deutsche Bibliothek — Cataloging-in-Publication Data
Bache, Carl:
Mastering English : an advanced grammar for non-native and
native speakers / Carl Bache ; Niels Davidsen-Nielsen. - Berlin ;
New York : Mouton de Gruyter, 1997
(Topics in English linguistics ; 22)
ISBN 3-11-015535-4
NE: GT
© Copyright 1997 by Walter de Gruyter & Co., D-10785 Berlin

All rights reserved, including those of translation into foreign languages. No part of this book
may be reproduced or transmitted in any form or by any means, electronic or mechanical,
including photocopy, recording, or any information storage and retrieval system, without
permission in writing from the publisher.
Printing: WB-Druck GmbH, Rieden am Forggensee.
Binding: Lüderitz & Bauer, Berlin.
Printed in Germany.
Acknowledgements
In 1993, the Danish Research Council for the Humanities launched five
foreign-language grammar projects. In being commissioned to write the
grammar of English in this programme, we were largely relieved from our
teaching duties. In addition to being sponsored by the research council,
this book has received financial support from Odense University (on the
recommendation of Fritz Larsen and Carl-Erik Lindberg), the Copen-
hagen Business School, Ingeni0r N. M. Knudsens Fond and from Ib and
Elise Bache. For all this generous support we would like to offer our sin-
cere thanks.
Peter Harder (Copenhagen University) and Fritz Larsen (Odense Uni-
versity) have been our readers during the entire project and have provided
us with extremely helpful and perceptive criticism. We are also grateful to
Alex Klinge (Copenhagen Business School), Leo Hoye (Odense Univer-
sity) and Henning Kirkegaard (Odense University) for proposing a number
of highly constructive revisions. Without the liberal assistance of these
five scholars the task of completing this book would have been harder and
the result poorer. Needless to say none of these persons can in any way be
held responsible for remaining errors and obscurities in the text.
We would also like to thank our colleagues in the parallel grammar pro-
jects on French, German, Italian and Russian for commenting critically on
selected topics in our grammar, presented at a series of very inspiring
cross-linguistic symposia.
In our project we have been fortunate enough to have Christian Heyde
Petersen (Odense University) as our research assistant. One of his various
tasks was to compile the subject index. We would also like to thank Stefan
B. Andersen (Odense University Press), Kjeld V. S0rensen, J0rn Erik
Wennerstrem and Elsebeth Jensen (Odense University Computer Service
Centre) for technical assistance. For assistance received in the late phases
of the work our thanks go to Anne Marie K011gaard (Copenhagen Busi-
ness School) for proofreading the entire manuscript and completing the
subject index, and to Eva Bang and Mimi Swiatecka (students at the Co-
penhagen Business School) for helping out with the word index.
Our grammar has been tried out on students of English at Odense Uni-
versity and at the Copenhagen Business School, not only by ourselves but
also by Alex Klinge, Christian Heyde Petersen and Marianne Stelen. Our
thanks are due to these colleagues and to our students (particularly Stefan
Mogensen) for providing a great deal of valuable feedback.
vi Acknowledgements
When in doubt about the acceptability of our examples we have conferred

with native speakers of English, in particular John Murphy (Copenhagen
Business School), John Dienhart, Leo Hoye, Sharon Millar, Tom Pettitt
(Odense University) and Penny Rosier and Phil Staines (Sydney). Our
thanks go to these colleagues as well for responding orally - with patience
and apparent interest - to questions posed over a long period, often at in-
convenient hours.
Copenhagen and Odense, April 1997
Carl Bache & Niels Davidsen-Nielsen

Contents
Parti
Chapter 1: Preliminaries 1
1.1. Goals and framework 1
1.2. Organization 1
1.3. Data 2
1.4. Varieties of English 3
1.5. Standard English 4
1.6. Grammatical variation 5
1.7. Variation according to medium 7
1.8. English for Special Purposes 11
Chapter 2: An introduction to syntax 13

2.1. The word 13
2.2. The sentence 15
2.3. Grammatical structure 18
2.4. Linearity and the principle of proximity 19
2.5. Constituency 20
2.6. Linguistic creativity and ambiguity 22
2.7. Competence and performance 24
2.8. Syntagmatic and paradigmatic relations 25
2.9. Recapitulation 28
Chapter 3: Elementary sentence analysis 29

3.1. The basic form and function approach 29
3.1.1. Form and function 29
3.1.2. Sentence functions 29
3.1.3. Four form types 31
3.1.4. Word classes 32
3.1.5. Simple complete analyses 36
3.1.6. Discontinuity 37
3.2. Sentence functions and sentence structures 38
3.2.1. The predicator 38
3.2.2. The subject 41
3.2.3. Basic sentence structures 44
3.2.4. More sentence structures 46
3.2.5. The direct object 47
viii Contents
3.2.6. The indirect object 50

3.2.7. The subject complement 52
3.2.8. The object complement 54
3.2.9. Adverbials 55
3.3. The internal structure of complex forms 59
3.3.1. The group 59
3.3.2. The preposition group 63
3.3.3. The compound unit 66
3.3.4. The clause 68
3.3.5. Embedding 69
Chapter 4: Advanced sentence analysis 71

4.1. Stacking 71
4.1.1. Form stacks 71
4.1.2. Function stacks: predicates and predications 73
4.1.3. A final comment on stacking 77
4.2. Missing constituents, ellipsis and pro-forms 78
4.2.1. The zero convention 78
4.2.2. Types of ellipsis 81
4.2.3. Pro-forms 83
4.3. Complex predicators 85
4.3.1. Phrasal verbs 85
4.3.2. Prepositional verbs 88
4.3.3. Phrasal-prepositional verbs 93
4.3.4. Other complex predicators 95
4.4. The top of the tree 97
4.4.1. Exit Sent 97
4.4.2. Communicative functions 98
4.4.3. The forms of communicative functions 102
4.4.4. Non-declarative clauses 104
4.4.5. Block language 105
4.5. Vocatives, interjections and dislocation 106
4.6. A final word on advanced sentence analysis 110
Part II
Chapter5: Constituent order 111
5.1. Introduction 111
5.2. Functions of constituent order 112
5.2.1. Grammatical relations 112
5.2.2. Illocutionary value 112
Contents ix
5.2.3. Information structure 113

5.3. Inversion 116
5.3.1. Preliminaries 116
5.3.2. Full and partial subject-predicator inversion 116
5.3.3. Partial inversion with illocutionary value 118
5.3.4. Partial inversion caused by an initial constituent 120
5.3.5. Full inversion and information structure 123
5.3.6. Full inversion after a fronted adverbial 124
5.3.7. Full inversion after a fronted participial predication stack 125
5.3.8. Full inversion after a fronted subject complement 127
5.3.9. Full inversion after a fronted object 127
5.3.10. Inversion of other sentence constituents 130
5.4. Constituent order in subordinate clauses 130
5.4.1. Relative clauses 131
5.4.2. Interrogative clauses 132
5.4.3. Exclamatory clauses 132
5.4.4. Other finite, non-finite and verbless clauses 133
5.5. Position and order of adverbiale 134
5.5.1. Main positions in finite clauses 134
5.5.2. Initial position 134
5.5.3. Medial position 135
5.5.4. Terminal position 137
5.5.5. Overview of positions in finite clauses 138
5.5.6. Positions in non-finite clauses 138
5.5.7. Factors governing the distribution of adverbiale 140
5.5.8. Position, form and relative weight 140
5.5.9. Position and (contextual cohesion 141
5.5.10. Position and scope 142
5.5.11. Position and semantic clarity 144
5.5.12. Position and style 145
5.5.13. Relative position of adverbiale 146
5.6. Discontinuity 148
5.6.1. Discontinuous clauses 149
5.6.2. Discontinuous verb groups 155
5.6.3. Discontinuous noun and pronoun groups 155
5.6.4. Discontinuous preposition groups 159
5.6.5. Discontinuous adjective and adverb groups 160
Chapter 6: Coordination and subordination 165

6.2. Coordination 167
χ Contents
6.2.1. Coordination and ellipsis 167

6.2.2. What can be coordinated? 169
6.2.3. Types of coordination 171
6.2.4. Coordinating conjunctions 174
6.3. Subordination 178
6.3.1. Introduction 178
6.3.2. Subordination at clause level 178
6.3.3. Subordination at group level 180
6.4. Markers of clausal subordination 183
6.4.1. Subordinating conjunctions 183
6.4.2. Interrogative and relative pronouns 185
6.4.3. Complex subordinating conjunctions 187
6.4.4. Correlative subordinators 188
6.4.5. Other markers of subordination 189
Chapter 7: The simple sentence 191

7.1. Introduction: simple and complex sentences 191
7.2. Referential properties: situations 191
7.2.1. Actionality: dynamic vs. Stative situations 191
7.2.2. Subtypes of dynamic situations 193
7.2.3. Subtypes of states 194
7.3. Participant roles 196
7.3.1. Introduction: general roles 196
7.3.2. Specific roles 197
7.3.3. Additional points 199
7.3.4. Formal links 200
7.4. Voice 205
7.4.2. Functions of the passive 206
7.4.3. Extended use of the passive 208
7.4.4. Voice restrictions 210
7.4.5. Nonfinite passives, GET-passives and notional 'passives' 214
7.4.6. Passives vs. adjectival non-passives 217
7.5. Polarity 219
7.5.2. Standard negation 220
7.5.3. Domain of negation 221
7.5.4. Syntactic field of negation 223
7.5.5. Negative sentences with global domain and limited field 224
7.5.6. Local negation 227
7.5.7. Nonassertive forms 228
Contents xi
7.5.8. Semantic scope of negation 230

7.5.9. Non-declarative sentences 234
7.5.10. Emphasis and focus 235
7.6. Subject-predicator concord 238
7.6.2. The basic rule 238
7.6.3. Singular or plural subject realization? 240
7.6.4. Notional concord 245
7.6.5. Attraction 249
7.7. Other types of external concord 250
Chapter 8: The complex sentence 253

8.2. Classification of subclauses 254
8.3. Subject clauses 256
8.4. Cleft sentences 258
8.5. Object clauses 263
8.6. Subject complement clauses 265
8.7. Indirect object clauses and object complement clauses 266
8.8. Adverbial clauses 266
8.9. Conditional clauses 269
8.10. Clausally realized disjuncts 271
8.11. Polarity in complex sentences 275
Part III
Chapter 9: Verbals 277
9.2. Verb forms 278
9.3. The external relations of verbals 280
9.4. The internal structure of verb groups 281
9.5. Auxiliaries and their delimitation 282
9.6. Tense and aspect 285
9.6.2. Deictic forms: present and past 287
9.6.3. Future forms 290
9.6.4. Perfect forms 293
9.6.5. Future perfect forms 296
9.6.6. Progressive forms: introduction 298
9.6.7. Present and past progressive forms 302
9.6.8. Future progressive forms 304
xii Contents
9.6.9. Perfect progressive forms 305

9.6.10. Future perfect progressive forms 306
9.6.11. Nonfinite progressive forms 306
9.6.12. Recapitulation 307
9.6.13. Tense-aspect in indirect speech 311
9.6.14. Tense-aspect in literary narrative 314
9.7. Modal uses of tense-aspect forms 316
9.8. Mood 319
9.8.2. The imperative 320
9.8.3. The subjunctive 322
9.9. Modality 324
9.9.2. MAY/MIGHT 327
9.9.3. CAN/COULD 329
9.9.4. MUST 332
9.9.5. SHALLfcHOULD 336
9.9.6. WILL/WOULD 339
Chapter 10: Nominals 341

10.1. Preliminaries 341
10.1.1. Nouns and noun groups 341
10.1.2. The external relations of nominals 341
10.1.3. The internal structure of noun groups 342
10.1.4. The functional domain of nominals 344
10.2. Categorization 350
10.2.1. What's in a head? 350
10.2.2. The semantics of nouns 352
10.2.3. Gender 353
10.2.4. Types of nouns and referents 355
10.2.5. Countability 356
10.2.6. Recapitulation 359
10.3. Determination 360
10.3.1. Types of determiner 360
10.3.2. Co-occurring determiners: pre- and post-determiners 362
10.3.3. Determiners and quantifiers 363
10.3.4. Referential orientation 365
10.3.5. The definite article 367
10.3.6. The indefinite article 372
10.3.7. Zero determination 378
10.3.8. The genitive 382
Contents xiii
10.4. Quantification: the number category 390

10.4.1. The regular singular/plural distinction 390
10.4.2. Irregular plurals 391
10.4.3. Number-invariable nouns 394
10.4.4. Collective nouns 395
10.4.5. What is pluralized? 396
10.4.6. The uses of the singular and the plural 397
Chapter 11: Pronominale 399

11.1. Preliminaries 399
11.1.1. Definition of the pronoun group 399
11.1.2. Classification of pronouns 400
11.1.3. The external relations of pronominals 401
11.1.4. The functional domain of pronominals 402
11.2. Central pronouns 404
11.2.1. Personal pronouns 404
11.2.2. Possessive pronouns 412
11.2.3. Reflexive pronouns 415
11.3. Pronouns without a person distinction 418
11.3.1. Demonstrative pronouns 418
11.3.2. Interrogative pronouns 424
11.3.3. Relative pronouns 428
11.3.4. Indefinite pronouns 436
Chapter 12: Adjectivals and adverbals 447

12.1. Preliminary discussion of adjectivals 447
12.1.1. Adjectives and adjective groups 447
12.1.2. Semantics 449
12.1.3. Morphology 451
12.1.4. The external relations of adjectivals 452
12.1.5. Parataxis and hypotaxis 455
12.1.6. Descriptive, classifying and specifying adjectives 456
12.1.7. Positional restrictions 458
12.2. Adjectival modification and positional ordering 462
12.2.1. The functional domain of adjectivals 462
12.2.2. Modificational zones 463
12.2.3. Inherent Mod. I, Mod. II and Mod. Ill adjectives 465
12.2.4. Structure in and across Mod. zones 466
12.2.5. Zone-internal order 469
12.3. Comparison of adjectives 473
xiv Contents
12.3.1. The basic system of comparison 473

12.3.2. Spelling and pronunciation 475
12.3.3. Irregular comparison 476
12.3.4. The choice between morphological and syntactic comparison 478
12.3.5. The use of compared forms 480
12.4. The substantival use of adjectives 483
12.4.1. What is meant by'substantival use1? 483
12.4.2. Generic and specific reference 484
12.5. Adverbals 486
12.5.1. Preliminaries 486
12.5.2. Morphology 488
12.5.3. VERY versus MUCH 491
12.5.4. The external relations of adverbals 492
12.5.5. Comparison 494
References 497
Subject index 503
Word index 525

PART I
1. Preliminaries
1.1. Goals and framework

This grammar is aimed at native and non-native students of English in
university and other tertiary education. At the same time, however, it is
sufficiently extensive, thorough and detailed to serve as a source of reference
for professional linguists and teachers of English. In order that it can be used
not only as an advanced textbook but also as a reference book, specific topics
are as far as possible dealt with separately and exhaustively. In this respect
the present book differs from a number of other recent grammars of English.
The grammar itself is not explicitly contrastive and therefore not limited to
a specific group of non-native students of English. It does, however, pay
special attention to characteristic features of English which are more acutely
felt by non-native than by native speakers of English, and in this sense it is
implicitly contrastive. For example, more attention is given to constituent
order - an important problem area for virtually all non-native speakers of
English - than in most other grammars of English of comparable size.
The present book is not written within the framework of any particular
linguistic theory (e.g. Functional Grammar or Chomskyan Generative Gram-
mar). Such a theoretical attachment would isolate us from too many readers.
In order to achieve the goals stated above we have been largely eclectic in
our descriptive approach, though not in a random fashion. Our guiding
principle will be a strict form/function distinction as it applies at all levels to
the constituents of the sentence. In this respect the syntactic framework
adopted here is largely the same as that presented in Bache, Davenport,
Dienhart and Larsen 1993. The structure of the present grammar is thus
determined by considerations of how to describe forms and functions most
appropriately at all constituent levels in English.
1.2. Organization
The description of English grammar given in this book is divided into three
parts. In Part I we offer an introduction to syntax (chapter 2), present the
descriptive framework, i.e. the form/function distinction at all levels of
analysis from sentence to word (chapter 3), and develop the sentence analysis
system to cope with complex syntactic issues, such as stacking, ellipsis, zero
2 Preliminaries
constituents, complex predicators and the relationship between sentence type

and pragmatic utterance function (chapter 4).
Part II describes basic syntactic characteristics in English: constituent
order (including inversion, discontinuity and the position and order of
optional adverbials) in chapter 5; coordination and subordination (including a
discussion of determination, complementation and modification) in chapter
6; the simple sentence (including a discussion of situation types and
participant roles, voice, negation and concord) in chapter 7; and the complex
sentence (including a formal and functional classification of subordinate
clauses) in chapter 8.
Part HI is devoted to group structure and word classes. It discusses verbs
and verb groups (chapter 9), nouns and noun groups (chapter 10), pronouns
and pronoun groups (chapter 11), adjectives/adverbs and adjective/adverb
groups (chapter 12). In this part of the book categories such as gender,
number, tense, aspect, mood and comparison are investigated.
Though the book is structured according to form, it is characterized by a
strong element of function. A great deal of attention will be given to syn-
tactic functions such as subject, object, predicator, etc. and head-dependent
relationships, to semantic functions such as agent, affected, instrument, etc.
and to pragmatic functions such as topic and comment. However, our book is
not a grammar of functions in the sense that it selects as its point of departure
a number of major communicative functions such as 'referring to people and
things', 'giving information about people and things', 'expressing time' and
'expressing manner and place' (cf. e.g. Collins Cobuild English Grammar
from 1990 and Downing & Locke 1992).
Consideration will also be given to information structure, and in this
connection the role played by stress and intonation and by textual factors
outside the sentence influencing its syntax will be taken into account. As a
very important function of the ordering of sentence constituents is to signal
the way in which a message is organized into information units, information
structure plays an important role in our account of constituent order in
chapter 5 and of voice in section 7.4.
1.3. Data
The approach taken to data in this book is non-positivist and instrumental.
We thus regard data as a means rather than an end: a means to secure
analytical breadth and precision as well as illustrative exemplification. Our
approach to grammar is not corpus-driven, and we do not see it as our task to
provide an exhaustive description of one or more corpora. Our examples are
derived from a number of sources: from modern British and American
Varieties of English 3
written texts (newspapers, magazines, fiction, etc.), from other grammars of

English (including descriptions based on corpora of spoken and written
English), from dictionaries and from introspection. Some of our examples
are thus not 'real'. In those cases where we have invented examples, we have
used our own intuitions about acceptability. When in doubt, however, we
have conferred with native speakers of English. Sometimes we have also
found it useful to modify authentic examples, thereby producing semi-
authentic examples. In those cases where our examples are from dictionaries
or from other books on English grammar we have indicated their source. In
the remaining cases we have not regarded the benefit and interest to the
reader of being informed about the source of each individual example as
sufficiently great to necessitate an indication of sources (in the text and in the
bibliography), which is both time- and space-consuming. Perceptive readers
will recognize the following authors as being among our favourite literary
sources: Martin Amis, Julian Barnes, A. S. Byatt, Len Deighton, John Irving,
P. D. James, John le Carre, Timothy Mo and Isaac Singer.
The view held on data in this book is that the examples should be the ones
that are relevant to a description of English grammar in all its aspects and
that it does not matter whether they are derived from corpora, authentic texts,
elicitation or introspection as long as they are relevant. Although corpora
constitute a highly valuable source of material, and although access to
corpora has been shown to reveal aspects of English grammar which have
not been captured before, we do not think that a description of English
grammar should proceed from authentic examples exclusively, for in that
case relevant data may be excluded.
1.4. Varieties of English

All languages are characterized by variation. Some varieties are user-related
and associated with language users living in a particular region or belonging
to a particular class. But besides such regional and social dialects there are
also varieties which are associated with special functions and which are use-
related. Such variation is to do with field of discourse (e.g. law, business,
science) or communicative situation (e.g. formal or informal) and differs
from regional and social variation in being transient. Under use-related
variation we can also include the difference between spoken and written
language. There are other types of variation as well, for example according to
sex and age, but the major types of variation may be said to be determined by
region, social group, style (function, situation/participants) and medium
(spoken/written).
4 Preliminaries
English is the most widely used language in the world. It is used by at least
750 million people in addition to being the mother tongue of about 350
million people. In countries like India, Nigeria, Kenya and Singapore it is a
second language, and is used for administration, education and broadcasting.
It is therefore hardly surprising that it is characterized by a great deal of
variation. Today its regional varieties differ from each other primarily with
respect to pronunciation and vocabulary. While Australian English, for
example, can be identified by a collection of pronunciation features (one of
which concerns the pronunciation of the diphthong in words like Australia
and mate as /aei/) and specific words and word meanings (for example red-
back, a particular kind of spider, and scrub 'poor vegetation'), its grammar is
remarkably similar to that of other regional varieties, particularly British
English. Variation according to field of discourse primarily is to do with
vocabulary. Legal English, for example, makes use of special legal terms and
archaic expressions such as aforesaid, aforementioned and hereinafter. On
the other hand its syntax, though tending to be rather complex, does not
differ significantly from that of other varieties. Predominantly social
varieties such as Cockney (the English used by working-class Londoners)
and Black English (used by some US citizens of African background) have
many special features of pronunciation and also many special words. But
here there are also several grammatical features which are not shared by
other varieties of English.
1.5. Standard English

The term 'Standard English' is widely used but is by no means easy to define.
While it may be difficult to speak of a standard language as a whole, it is
much less problematic to characterize some specific features as standard (e.g.
the ending -s in the 3rd person singular form of English verbs: she wants)
and other specific features as non-standard (e.g. the absence of the -s ending
in such a form). A large number of those features which are considered
standard in English today are derived from the Middle English dialect spoken
in the East Midlands which became predominant as the official form of the
language and was therefore the one preferred in writing and printing.
'Standard English' is used to describe the variety which is today most
widely accepted and understood either within an English-speaking country
(for example Standard American English) or throughout the English-
speaking world (Standard General English, understood as a supra-regional
language, or 'standard of standards'). Linguists tend to agree that Standard
English is most easily identified in print (irrespective of pronunciation,
which varies considerably from place to place), that it is the variety used by
Grammatical variation 5
most newsreaders on radio and television networks (BBC, CBS, NBC, ABC,
CBC, etc.) and that it relates to social class and level of education (see
Me Arthur 1992, on which sections 1.4, 1.5 and 1.6 of this chapter are largely
based). It is the written form used by all educated British writers in neutral or
formal style. As Standard British English is remarkably similar to that of
other national standards, for example the American, Australian and Canadian
standards, it has been claimed to be the written form used by writers of
English throughout the world. It is the English we find, for example, in the
New York Times, the Independent, the Toronto Globe and Mail and the
Sydney Morning Herald and which is described in Webster's Third New
International Dictionary and the Oxford English Dictionary. This
'monocentric' view, according to which English has a (British) core and a
periphery, has been challenged by some scholars, however, who hold that
English has become 'pluricentric' and that it is therefore more correct to
speak of'Englishes' than of'English'.
In this book we describe the grammar of Standard British English (BrE).
Owing to the grammatical similarity between the different national standards
this description will apply very largely to other national standards as well. As
American English (AmE) is particularly important with respect to range and
number of speakers, we shall, however, account for specific differences
between BrE and AmE in the course of our description. We shall also pay
attention to the difference between spoken and written English and describe
grammatical features which are characteristic of spoken English.
1.6. Grammatical variation

The most important regional varieties of English are American, Australian,
British, Canadian, Caribbean, Indian, Irish, New Zealand, Scottish and South
African English. Regional subvarieties occur as well, as exemplified by
Northern British English and by the American dialects spoken in Eastern
New England and the South. Social subvarieties of a regional variety can be
illustrated by the occurrence in Indian English of three levels: the acrolect
(educated Indian English), the basilect (pidginized varieties) and the
intermediate mesolect. Among the predominantly social varieties Cockney,
Black English and Chicano English (the English used by speakers of
Mexican heritage in the US Southwest) are particularly noteworthy. As
pointed out in section 1.5, the grammatical differences between these
regional and social varieties will not be systematically accounted for in this
book. What we shall do instead here is to illustrate grammatical variation by
looking briefly at a few selected grammatical features (all of which will be
dealt with more extensively in subsequent chapters).
6 Preliminaries
In BrE the present perfect (e.g. has signed) and the simple past (e.g. signed)
are both used to describe events that precede the moment of speaking, but
they differ in perspective. The former presents a past time event as having
implications about what is true of the moment of speech, e.g. She has signed
the letter. The latter presents a past event as something which has no such
implications and which is in this sense over and done with, e.g. She signed
the letter. The two verb forms are used in basically the same way in e.g.
Australian, New Zealand, Scottish and South African English as in BrE. In
AmE, however, the simple past is often used where the present perfect is
used in other varieties, for example - addressing a child about to go to bed -
in Did you brush your teeth?, where a very recent past event is referred to.
Other examples illustrating this American use of the past are Did the
children come home yet? and You already told me. In Indian English (the
mesolect), on the other hand, the present perfect is often used under
conditions where the simple past is used in other varieties of English, for
example in a sentence like He has bought the car yesterday.
The so-called progressive construction - which consists of a form of be
followed by a present participle, as in She was reading - is used to present an
event as an ongoing process. In most varieties of English this construction is
ruled out if the verb is meant to be 'stative', i.e. meant to express a relational
state of affairs (e.g. contain) or inactive perception or cognition (e.g. hear,
know). In some varieties, however, there are fewer restrictions on what verbs
can be used in the progressive. In Scottish English we find progressive
examples with stative verbs like He was thinking he'd get paid twice and in
Indian English (the mesolect) examples like Lily is having two books (these
examples and the ones given below in this section are quoted from McArthur
1992). In other varieties of English, progressive meaning is reinforced by a
special word: in Black English this is steady, as illustrated by We be steady
rappin', and in South African English it is busy, as in We were busy waiting
for him. In Black English progressive meaning can also be expressed by
steady exclusively, as in They be high steady.
Thirdly, some varieties differ from others with respect to modal verbs. In
AmE, shall and ought are rarely used outside formal style. In Scottish
English there are several nonconformist uses of the modals: shall and may
tend not to be used in informal speech, must is not used to express
compulsion (for which have to or have got to are used) and need and dare are
not used as auxiliaries but as main verbs exclusively, as in He didn't need to
do that (not He needn't do that) and She doesn't dare to talk back (not She
daren't talk back). This last property is shared by Black English as well.
As appears from these examples, many varieties of English stand out in
that they differ grammatically from Standard British English.
Variation according to medium 7
1.7. Variation according to medium

There is in many languages a good deal of variation according to medium:
the grammar of the written variety often differs significantly from that of the
spoken variety. The extent to which speech and writing differ varies from
language to language: in e.g. Arabic there are different standardized varieties
for the two media and the relationship between them is tenuous (a situation
referred to as 'diglossia'), while in English speech and writing are felt to be
predominantly stylistic variants.
Before describing some of the characteristics of these variants, it is import-
ant to make a distinction between medium and channel of communication (cf.
Lyons 1981: 18). Speech and writing are best understood as different media
serving different communicative purposes in different contexts, as distinct
from the actual (oral or written) channel of communication. Thus speech is
used in e.g. everyday conversation while writing is used in e.g. newspapers.
Usually writing is delivered 'in writing' and speech is actually spoken. But
this is not necessarily so: we can speak the way we write (e.g. when we read
aloud), and we can write, more or less, the way we speak (e.g. in e-mail and
in dialogues in a novel). Despite such cases where we mix medium and
channel of communication, the fact remains that speech and writing are often
structurally and functionally different varieties of language, each facilitated
(and restricted) by its usual channel of communication. Thus, for example, in
speech we may rely heavily on prosody (intonation and stress) and para-
linguistic means (such as e.g. voice quality, gestures, eye-contact, smiles,
frowns, yawns, etc.) to get our message across. In writing, we are left with a
number of conventional symbols for organizing our message: full stops,
commas, question marks, exclamation marks, bold face, capitals, etc., not to
mention the choice between handwriting and typing/printing. On the other
hand, speech is more transient (unless, of course, someone makes a specific
point of recording it), and it is more difficult to edit than written language,
which is usually the product of relatively careful planning and drafting, and
which always leaves you with a record of the communicative event. The
communicative interaction between the participants of a speech situation is
more immediate and complex than when writing is involved.
In terms of more specifically grammatical features, spoken and written
English differ in a number of important ways. Characteristically, writing,
unlike speech, is lexically very economical and, at the same time, dense:
written sentences generally contain fewer (partially or wholly) 'redundant'
words but more 'heavily loaded' words than spoken utterances. A feature that
adds to the density of writing is the greater complexity of some of the units
that make up the sentence, especially the so-called noun groups (cf. section
8 Preliminaries
3.3.1), which tend to be longer and to contain more levels of structure than in
speech. Conversely, speech has greater density or complexity in the
organization of clauses (in terms of what is usually referred to as
'coordination1 and 'subordination1, cf. sections 3.3.3 to 3.3.5). These differ-
ences are borne out very nicely by examples like the following (from
Halliday 1985) where the a-variants are typical instances of written language
and the b-variants of spoken language:
(la) Investment in a rail facility implies a long-term commitment.
(Ib) If you invest in a rail facility, this implies that you are going to be committed
for a long term.
(2a) The growth of attachment between infant and mother signals the first step in
the development of a child's capacity to discriminate amongst people.
(2b) When an infant and its mother start to grow attached to each other, this is a
sign that the child is beginning to discriminate amongst people.
Note also that the a-examples of writing are fairly static in their presentation
of the message, whereas the b-examples of speech are much more dynamic.
When information is crammed into a few complex units (as in (2a)), what we
refer to becomes rather fixed, factual and unchanging. But when basically the
same information is spread over a number of clauses (as in (2b)), we get a
clearer sense of the activities and processes involved: there are more verbs
and consequently a clearer time sequence emerges.
Apart from the very general differences between speech and writing
mentioned so far, there are numerous more specific differences pertaining to
the grammar of English. Let us mention a few of these. In written English, an
adverbial (i.e. a sentence function which is not a subject, predicator, object or
complement, see section 3.2.9) is frequently realized as an -ing participle
clause or an -ed participle clause:
(3) Giving him a light, I set fire to his moustache.
(4) Her oration finished, she breathed heavily with an overflow of indignation.
While this type of realization is not an exclusive property of the written
medium, it is found less frequently in spoken than in written English.
Secondly, the so-called subjunctive is typical of written English:
(5) Whatever be the reason, we cannot tolerate his disloyalty.
(6) Grafton would have rung if the plane weren't on its way.
In informal spoken BrE, these subjunctive forms would be replaced by may
be (placed after reason) in (5) and by wasn't in (6).
In sentences like We must put some flesh on your bones and I just saw a
show on television where there is both an object (some flesh and a show, re-
spectively) and an adverbial (on your bones and on television, respectively),
Variation according to medium 9
the former typically precedes the latter. Sometimes, however, this ordering
may be reversed (as we shall return to in section 5.3.10). This can be
illustrated by the following examples in which the adverbial is placed before
the object:
(7) I just saw [on television] [how some Indian people started a shop and put the
old grocery on the comer out of business].
(8) Hello, my name is Penny Rogers. I bought [some time ago] [a PowerBook
180]. I can't get the internal modem to work and would like to have someone
look at it.
This ordering is found in both writing and speech, but often for different
reasons: in writing it is the result of careful planning and involves con-
sideration of e.g. weight as in example (7), where the object is very long and
hence preferred at the end of the sentence to prevent it from unduly delaying
the occurrence of the adverbial. In speech the ordering of the adverbial
before the object is often the result of lack of planning, or rather, planning on
the spur of the moment: the order of the units here reflects the order in which
the speaker thinks of what to say rather than any strict grammatical principle.
As a final example of a grammatical feature which is typical of written
English, it is the case that in English the verb, or 'predicator', may be placed
before the subject if an adverbial is fronted to give prominence to it or to
establish narrative continuity (see section 5.3.6 on so-called full inversion
after a fronted adverbial):
(9) On the walls were pictures of half-naked women and colourful landscapes.
(10) On the doorstep sat women nursing their babies and gossiping.
This ordering is virtually only found in written English. In spoken English
there would be used as a 'provisional subject1 in (9), and in (10) the
predicator sat would be placed after the subject.
In spoken English we find utterances of the following kind:
(11) What a load of rubbish!
(12) Mind if I smoke?
Such 'elliptical' constructions and 'non-sentences' abound in both speech and
writing but often for different reasons. In speech, the dropping of redundant
words or constructions is the result of a reliance on the immediate context
and part of the easy-going flow of the conversation and the smooth turn-
taking of the participants. In writing, such 'telegraphic style' is used to catch
the receiver's attention (e.g. headlines, road-signs, chapter headings, titles,
warnings, neon commercials) or to arrange the message in a clear,
systematic, comprehensible manner (timetables, recipes, shopping lists, bank
statements, television programmes, sports results, etc.). In writing, unlike
10 Preliminaries
speech, catching the receiver's attention cannot be done prosodically or

through the use of gesture, etc.
Another characteristic feature of spoken English is the frequent occurrence
of so-called 'comment clauses' - i.e. clauses like you know, I take it,
generally speaking and to be honest which serve to add a parenthetic
comment to another clause, as in / don't think you'll pass, to be honest. In
speech it is especially stereotyped comment clauses like the ones just
mentioned that are frequent. Comment clauses which are less idiomatic, and
which require more planning, for example (This is a serious mistake,) he will
undoubtedly have realized by now, are not typical of the spoken medium.
So-called conditional utterances are typically marked by if or unless, as
illustrated by If you do that again I'll strangle you and Unless you shut up I'll
strangle you. But they can also be expressed by an imperative construction
followed by a statement introduced by and or or:
(13) Do that again and I'll strangle you.
(14) Shut up or I'll strangle you.
What the speaker expresses in these examples is an intention to inflict injury
on the addressee if a certain behaviour continues or unless a certain
behaviour is discontinued. This way of signalling a conditional threat is
typically restricted to spoken English.
As a final illustration of a grammatical feature which is characteristic of
spoken English, we can mention examples of 'dislocation' (cf. section 4.5):
(15) He's an utter nitwit, that boyfriend of yours.
(16) Your brother George, I've never understood why he didn't resign.
Here the identity of the person referred to is established by a noun group
which is either added as an amplifying tag (as in (15)) or prefixed to the
sentence (as in (16)).
In closing this section we should point out that differences between spoken
and written English like the ones illustrated in this section are also largely
characteristic of formal vs. informal English, whether written or spoken.
Such differences thus characterize not only variation according to medium
but also variation according to style. Many of the grammatical features which
are typical of spoken English are found also in informal written English, for
example private letters or memos. Conversely, many of the features
characterizing written English are found also in so-called edited speech, for
example lectures and political speeches, where medium and channel of
communication are conventionally mixed.
English for Special Purposes 11
1.8. English for Special Purposes

When describing variation it is customary to distinguish between Language
for General Purposes (LGP) and Language for Special Purposes (LSP). LSP
refers to varieties used by practitioners of a profession in their work (see
Kragh 1991). As pointed out in section 1.4, variation according to field of
discourse (law, business, science, technology, etc.) is use-related in the sense
that it involves switching to a variety which the occasion demands. But as a
variety of this sort is typically used by practitioners of a special profession,
i.e. by specialists who have gone through a professional socialization process
which is partly linguistic, it is in fact user-related as well. Legal language is
typically used by members of the legal profession, scientific language by
scientists, economic language by economists, and so on.
LSP is primarily characterized by its vocabulary, i.e. by special terms
employed by a profession, such as lien, liability, habeas corpus, statutory
and aforesaid in legal English. But it is also characterized by features of
grammar which are particularly frequent. While some of these are typical of
formal (vs. informal) and written (vs. spoken) English as well, others are
largely restricted to a special professional variant.
English for Special Purposes (ESP) tends to be rich in complex noun
groups. This can be illustrated by groups like the FT-SE 100 Index of shares
in Britain 's leading companies (business) and the issues of breach of
statutory duty and common law negligence in respect of the council's
exercise of its power under the Act (law). When realizing an object, such
complex noun groups will often necessitate the reversal with an adverbial
mentioned in section 1.7. In the following example the adverbial on local
authorities in the clause beginning with to has been moved forward because
of the length of the object noun group (the part of the sentence stretching
from any to statute):
(1) There was a considerable reluctance on the part of the courts to impose [on
local authorities] [any liability for breach of statutory duty other than that
expressly imposed in the statute].
The complexity of noun groups is often due to a string of words occurring
before the noun constituting the nucleus, or head, of the construction. In its
frequent use of such heavy premodification in noun groups ESP differs from
English for General Purposes (EGP). In scientific English we find noun
groups like inertial confinement fusion, near-zero explosive yields and the
first full digital image model of Mars, in which the head nouns are fusion,
yields and model. In business English heavy premodification can be
illustrated by global gross domestic product, International Business
Machines' year-end results and purchasing power parity exchange rates,
12 Preliminaries
where the head nouns are product, results and rates. In legal English we find
heavily premodified nouns too, for example in the Nurseries and Child
Minders Regulation Act 1948. Owing to its general condensation and the
way in which its noun groups tend to be packed with information, LSP is
sometimes informally referred to as 'agglomerese'.
As LSP is typically used to describe and direct, whereas emotive and social
uses are not normally involved, it must aim at being clear, concise, objective
and reliable. Such pragmatic requirements affect grammatical choices.
Descriptive and directive technical texts, for example, have been shown to
contain many passive constructions where the preferred verb form is in the
present, and many compound nouns and adjectives which have been derived
from clauses (see Munck 1991). In English this can be illustrated by an
example like the following (quoted from Me Arthur 1992: 1026):
(2) Three modes of operation are required: voice-activated mode (VOX), press-
to-talk (PTT) and call.
Here the choice of the passive contributes to making the message objective
(impersonal), its present form to making it general (what is described is valid
at all times), and the use of compounds to making it concise. Conversion of
clauses into compound words for the sake of brevity can be further illustrated
by examples like quick-drying (ink), quick-action (reversing gear), rapid-
hardening (cement), diesel-powered (engine) and self-raising (flour). It can
also be noted that in order to avoid ambiguity, the second of two noun groups
referring to the same entity is sometimes not replaced by a pronoun as it
typically is in EGP.
In legal English the modal verb shall is used with third-person subjects to
denote what is legally mandatory:
(3) The tenant shall quietly possess and enjoy the premises during the tenancy
without any interference from the landlord.
This usage is not found in current EGP, nor in other types of ESP. While
there are thus features of grammar which are restricted to (a variety of) ESP,
the grammatical differences between ESP and EGP are nearly always
quantitative rather than qualitative. We find the same features of grammar in
both varieties, but the frequency with which they occur is often markedly
different. As we saw in section 1.7, this is also largely the case with varieties
engendered by differences in medium.
2. An introduction to syntax
2.1. The word

As native speakers of a language, and very often also as learners of a foreign
language, we have an intuitive knowledge of that language, including its
syntax and the basic units of its grammar. Thus, for a start, we all have a
pretty good idea of what a word is. To realize this, we only have to consider
the following passage, where we have eliminated any indication of word
boundaries (such as, typically, an empty space between words):
(1) thepolicemanlookedatthembothhesnirTedthatwastheuncooperativeattitudeyou
mightexpectfromafamilythatencouragedtheirdaughtertogoaroundwithyanksan
dthesewerewelltodopeoplenotworkingclasssuchlaxattitudesoffendedhimhe'd
makesurethatnodaughterofhiskeptcompanywithforeignsoldiers
Once we recover from the initial confusion of having to decipher such a
muddled, uninviting passage, we can all find the individual words of the
original text:
(2) The policeman looked at them both. He sniffed. That was the uncooperative
attitude you might expect from a family that encouraged their daughter to go
around with Yanks. And these were well-to-do people, not working class.
Such lax attitudes offended him. He'd make sure mat no daughter of his kept
company with foreign soldiers.
There are, admittedly, occasional problems: is well-to-do one or three words?
Is He'd one or two words? But apart from such nitpicking, we are perfectly
capable of identifying the words of any language familiar to us. In writing,
word boundaries are signalled by blanks or by punctuation marks. In speech,
they are often, though not always, signalled by factors such as the exact onset
of stress (as in see the 'meat vs. see them 'eat) and/or the variant of speech
sound selected (as in keeps ticking vs. keep sticking, where the /t/ is aspirated
(i.e. pronounced with a puff of air) when it is a word-initial sound as in the
former case). We all know how to signal and interpret word boundaries in
both writing and speech, if only intuitively. And yet, amazingly, it is very
difficult to define what a word is.
Obviously, meaning is somehow involved: policeman means one thing,
family another. But what is the meaning of the and ofi Clearly, these words
mean something but their meaning is not as immediately transparent as the
meaning of policeman and family, which express relatively concrete entities
(more specifically, persons). And why is it that the and policeman are two
words in English but only one word in, for example, Danish (namely
14 An introduction to syntax
politimanden, where the ending -en corresponds to the English the)! Such
frivolity is not reserved for Danish and other foreign languages but is a
regular feature of English, too, as we see in the word uncooperative, where
un is only part of a word despite the fact that it has an independently
identifiable meaning. Another example is, once again, policeman: why is
policeman one word but both police force ana police constable two? Sim-
ilarly, why are there two separate words in class struggle but only one in
classroom, which is normally identified as a compound word (i.e. a unit of
elements which function independently elsewhere)? We are forced to
conclude that words cannot be defined simply as 'units of meaning'.
In the language user's conception of words, convention seems to be an all-
important factor. This, however, should not prevent us from trying to
describe the words in English (for which the technical term lexicon is often
used) with reference to any regular pattern applying to them. As a first step
towards such a description, grammarians refer to the smallest meaningful
units of language as morphemes whether or not they are independent words.
In this sense un- in uncooperative, police- and -man in policeman, work- and
-ing in working, and even -s in attitudes and -ed in encouraged are morph-
emes. The, him, of, to, that etc. are both morphemes and words. They arefree
morphemes in contrast to un-, -s, -ing, -ed, etc., which are bound morphemes.
This means that a word consists of one or more morphemes. The precise
identification of words is then to a large extent a question of conventional
rules of morphology, i.e. rules describing the structure of words in terms of
morphemes. There is little consistency across languages in the morphology
of words: as we have seen, the meaning of definiteness is in English typically
expressed by an independent word, the definite article the, whereas in Danish
it is typically expressed by a word-internal bound morpheme. Sometimes
principles seem to vary even within one and the same language (as in the
case of English policeman vs. police force). Note also that while definiteness
in English is expressed by means of an independent word, meanings
pertaining to, say, number (singular or plural) and tense (present or past) are
fully grammaticalized in that they are expressed by word-internal bound
morphemes, more specifically by inflections. Our intuitive knowledge of the
words of a language includes the knowledge of what is conventionally
expresssed by means of individual words and what is conventionally
grammaticalized at the morphological level.
One important characteristic of words is that they are basic syntactic units,
i.e. the building blocks of larger language constructions, and thus have a high
degree of stability ana cohesion. Words are stable in the sense that - unlike
many higher-level syntactic constructions such as the sentence - they do not
allow rearrangement of their constituent parts. Nor do they allow internal
The sentence 15
separation. For example, as language users we are not free to organize the
morphemes in words as we please. We have to say childishness, not
*nessishchild, *ishchildness, etc. With higher-level constructions there is
often a certain variability: we can say either the sickening unresolvable mess
or the unresolvable sickening mess (with little or no difference of meaning)
and we can say both Bob kissed Gina and Gina kissed Bob (although here
there is a marked difference of meaning). Note also that normally words are
internally inseparable. Thus while we are often free to separate independent
words like the and policeman in a construction like the policeman by
inserting an adjective, as in the young policeman, we cannot separate the
individual parts of the or policeman and still retain their status as single
words. In speech, words are also coherent in the sense that we can insert
pauses (uh, uhm, etc.) between words but not usually within words (cf.
Bolinger 1975: 119). Thus in an unsure and hesitant manner we might say:
(3a) The uhm policeman uh got uhm confused.
But we are unlikely to say:
(3b) The po-wA-liceman got con-wA-fused.
Stability and cohesion may be important clues in the identification of word
boundaries. But they tell us little about what a word really is or about why
the principles of word formation differ between languages and even within
languages.
The interesting fact is that despite the problem of formulating a water-tight
definition of the word, we all have an intuitive knowledge of what a word is.
That knowledge comprises in part an awareness of morphemes as units of
meaning, in part the recognition of largely conventional rules of how
morphemes combine to make up the units that we know as words.
2.2. The sentence

Grammar is not just the study of words and their morphological structure but
also of how words combine to make up larger units, such as sentences. Like
words, sentences are notoriously difficult to define rigidly and objectively.
And yet we all have an intuitive knowledge of what a sentence is (cf.
Bolinger 1975: 156). To appreciate this, we only have to look at a passage
where we have left out all the conventional markers of sentence boundaries
(such as punctuation and capitalization of initial letters after full stops):
(1) Victoria shuddered once again she realized that her father was trying to
protect her and she loved him for it and if she admitted to knowing Vince
Madigan the next question must inevitably be and what was this American's
relationship with Mrs Hardcastle and then more questions I don't recognize
him she said softly
Although this passage is a fairly complicated text with both internal and
external dialogue, it is easy to guess at its division into sentences:
(2) Victoria shuddered. Once again she realized that her father was trying to
protect her and she loved him for it. And if she admitted to knowing Vince
Madigan, the next question must inevitably be, 'And what was this
American's relationship with Mrs Hardcastle?' And then more questions. Ί
don't recognize him,' she said softly.
Many people will even discover that there are alternative ways of dividing
this text into sentences. Thus Once again could equally well belong to the
first sentence: Victoria shuddered once again. She realized... Similarly, and
she loved him for it might be a separate sentence: She realized that her father
was trying to protect her. And she loved him for it. But no-one would suggest
that Once again she is a separate sentence. Nor would we allow the long
sentence And if she admitted to knowing ... Airs Hardcastle? to be broken
into two independent sentences And if she admitted to knowing Vince
Madigan and The next question must.... Mrs Hardcastle, despite the fact that
both contain a verb. While the second part could conceivably function as a
sentence on its own, the first part is clearly incomplete. It cannot stand alone.
According to conventional wisdom, the two parts are clauses within the
same sentence.
In speech, clause and sentence boundaries are typically signalled and
interpreted in terms of tone groups ending with a special intonational
contour: e.g. a rise (to signal, say, the end of a question or the continuation
from one clause to another) or a fall (to signal, say, the end of a sentence). To
get a sense of such intonational signals, one can try reading out the passage
above with the different segmentations proposed.
Though we have to recognize And then more questions as an independent
unit, many would hesitate to call it a sentence. It is somehow unfinished,
lacking a verbal component such as (And then more questions) would follow.
Similar problems arise with short units of text like:
(3) No!
(4) After him.
(5) My turn?
Typically such units are complete utterances (cf. Bache et αϊ. 1993: 183ff).
But are they sentences? Although they are perfectly acceptable in both
speech and writing (in writing as a substitute for a spoken utterance), we
hesitate to accept them as sentences. It would help considerably if we treated
them as somehow short forms of'proper' sentences like the following:
The sentence 17
(6) I say no!

(7) I want you to go after him.
(8) Is it my turn?
But such 'full constructions' are often cumbersome and not entirely natural or
appropriate in context and therefore should not be taken as 'more proper' than
those in (3) to (5). We have to accept that not all utterances are sentences.
Many linguists make a systematic distinction between 'sentence' as a
theoretical unit (defined by grammar) and 'utterance' as a physical unit (a
matter of speech production), cf. e.g. Lyons 1995: 32ff. On this view some
utterances can be analysed in terms of sentences but utterances do not
'consist of sentences.
Some grammarians have suggested that a sentence is a unit of grammar
expressing a 'complete thought* or a 'complete event'. But surely such
definitions are too imprecise to be of much use. Is the thought or event
expressed in Victoria shuddered any more complete than those expressed by
examples (3) to (5), which are not full sentences? Or is it more complete than
that expressed by the non-sentential clause And if she admitted to knowing
Vince Madiganl Or than that expressed by And then more questions?
What all this amounts to is that although we have intuitions about what a
sentence is, and though we are perfectly capable of dividing a text into the
appropriate orthographical or intonational units typically reflecting sentence
or utterance boundaries, it is by no means obvious how actually to define a
sentence. We usually expect a sentence to contain at least a verbal compon-
ent and some other unit, but as we have seen, there are complications.
Sometimes textual units which do not meet this requirement are treated like
sentences in terms of punctuation or intonation, or in terms of their
independence as acts of communication. Furthermore, there is a complex
relationship between clause and sentence. Somehow, clauses are like
sentences in normally requiring a verbal component and some other unit. So
what is the difference between the two? In a sense the distinction between
clause and sentence is very similar to that between morpheme and word. A
sentence seems to consist of one or more clauses the way a word consists of
one or more morphemes. Sometimes a clause is also a sentence (and thus
resembles free morphemes which are independent words), sometimes a
clause is simply too incomplete or dependent to serve as a sentence in its
own right (and thus resembles bound morphemes, which never occur
independently). While it is possible to consider words to be the basic units of
syntax, the building blocks of larger units, sentences can be viewed as the
maximal autonomous units permitting syntactic analysis. Sentences and their
internal arrangement of words are the domain of syntax.
2.3. Grammatical structure

Our intuitions about language are not restricted to the mere identification of
possible words and sentences but include the organization of words within
sentences, i.e. the grammatical structure of sentences. Consider the follow-
ing example:
(1) John kissed the little old woman who owns that shaggy dog.
We doubt that our readers have ever seen a sentence completely identical to
(1). And yet no one has any difficulty in recognizing it as a grammatical
construction in English. In other words, there is an appropriate organization
of the eleven words in the sentence: they are all used in the right place, at the
right time. We know the individual words and their meaning, and somehow
we know the kind of relationship they enter into. To appreciate that this
organization of words, the structure of the sentence, is not random, we only
need to change the order of its constituent words:
(2) old the kissed dog shaggy who John woman little that owns.
Although we have exactly the same words here as in (1), (2) is completely
ungrammatical. It has become a list of unrelated words.
If we consider the possible ways of arranging the eleven different words of
(1) and (2) in a linear sequence, it is in fact a small miracle that - almost
without thinking about it - we hit on the grammatical sequence in (1). There
are, to be exact, 39,916,800 different ways of combining eleven different
elements in a sequence ( 1 x 2 x 3 x 4 x 5 x 6 x 7 x 8 x 9 x 1 0 x 1 1 ) . Some of
these many alternatives to (1) are of course perfectly grammatical:
(3) John kissed the old woman who owns that shaggy little dog.
(4) John kissed the little woman who owns that shaggy old dog.
(5) John kissed the woman who owns that shaggy little old dog.
Stretching our imagination a little we may even accept sequences like the
following:
(6) The shaggy little dog who owns that old woman kissed John.
(7) The old woman who owns little John kissed that shaggy dog.
(8) John owns the old woman who kissed that shaggy little dog.
These sequences are all grammatical (in the sense that the words enter
acceptable, recognizable syntactic relationships), but their meaning may
differ from our conception of what constitutes the normal state of affairs in
the world (Can a dog own a woman? Can a human being own another human
being?) and thus challenge us to think of contexts where it would be
appropriate to use such sentences.
Linearity and the principle of proximity 19
But even if we allow for a little stretching of our imagination, there are at the
very most, maybe about a hundred possible sentences containing the eleven
words in (1). There are millions of unacceptable ones. And yet we all have a
fairly impressive ability to spot the very few grammatical sentences and
reject all the ungrammatical sequences. This ability presupposes an intuitive
knowledge of the possible syntactic relationships between words. In other
words, we have an intuitive knowledge of grammatical structure.
2.4. Linearity and the principle of proximity

Let us have a closer look at our intuitive knowledge of grammatical
structure. As we have already seen, language is necessarily linear in the
sense that one constituent unit (a speech sound or a letter, a morpheme, a
word, a group of words, a clause, a sentence) always follows another. In
speech language takes time, and in writing it takes up space. Grammatical
structure is basically a means by which language comes to terms with, and
makes the best of, this basic condition.
Strictly from the point of view of linearity alone, we would expect a
sequence of elements to be either random, with no discernible patterns in the
organization of the elements, or progressively related, each element receiving
its rank according to its position in the list (in terms of, for example,
increasing or decreasing 'importance' or 'priority', or according to some
convention, such as 'alphabetical order"). In human language we see both
these main types of linear organization. A telephone directory is a good, if
fairly artificial, example of progressive linearity, and so is counting. But
alphabetical and numerical order is also exploited in many natural
expressions, such as:
(1) Gina got many A 's and B'sin her finals.
(2) You are a nice chap but you will have to watch your P's and Q's.
(3) Jim and Roger came in first and second, respectively.
(4) They arrived in twos and threes.
(5) He bought ten or twelve good books.
In a phrase like Ladies and Gentlemen, etiquette dictates a certain priority.
Progressive linearity is also present in constructions like:
(6) Gina is a competent, even brilliant, scientist.
which reflects an increase of the intensity with which Gina is described.
Random linearity may be present in constructions like the following:
(7a) A lex, Stephanie and Roger went sailing this morning.
(7b) Roger, Stephanie and Alex went sailing this morning.
(7c) Stephanie, Alex and Roger went sailing this morning.

etc.
(8a) She almost enjoyed the warm stale sweet air.
(8b) She almost enjoyed the warm sweet stale air.
(8c) She almost enjoyed the stale sweet warm air.
etc.
However, the basic randomness of the italicized constructions may be
reduced by considerations of rhythm or by contextual factors.
There is a different, more general, derived sense in which linearity is
important in the organization of language: since simultaneity of expression is
excluded, we can predict that, in compensation, elements that somehow
'belong together* will be placed as closely together in the sequence as
possible. Thus, in examples (7a-c), Alex, Roger and Stephanie belong
together (in that they all took part in the event expressed by the rest of the
sentence, i.e. they all went sailing) but since we cannot express them
simultaneously they are instead placed as closely together as possible. The
same applies to sweet, stale and warm in examples (8a-c): they belong
together because they perform the same function in the sentence, namely that
of describing air. As they cannot be expressed simultaneously but are forced
into a sequence, they are at least placed closely together. Given the condition
of linearity, it is thus in a sense natural that words that belong together
should be placed together in the sequence of words making up the sentence.
In this way we can say that the necessary linearity in the organization of
language leads to the principle of proximity.
2.5. Constituency
Grammatical structure imposes an organization on the elements of the string
which is neither progressive nor random. In doing this, grammatical structure
usually exploits the principle of proximity to create groupings of words that
belong together. Let us consider the following short version of example (1)
in section 2.3:
(1) John kissed the little old woman.
In this sentence, the word the is not in a random position relative to the other
words, nor does it receive any rank according to its place in a progression of
elements. Rather it is part of a grammatical structure in which it relates more
closely to woman than to John, kissed, little or old. At first blush the order of
words in this example seems to violate the principle of proximity: woman is
further away from the than John, kissed, little and old. But on closer
examination, it appears that the, little, old and woman all belong together in a
Constituency 21
group (according to the principle of proximity) and as such enter a 'joint'

relationship with kissed and John at a higher level. The sentence describes an
instance of kissing (expressed by kissed) in which there are two participants:
one who performs the kissing (John) and one who receives the kiss (the little
old woman). There is thus an indication that the sentence can be divided into
three parts or constituents: [John], [kissed] and [the little old woman]. The
interpretation of [the little old woman] as a group of words belonging
together is supported by the fact that if we want to move one of the words
relative to [John] and [kissed] and preserve the meaning of the three
individual parts of the sentence, we normally have to move them all:
(2) The little old woman kissed John.
(3) *Woman kissed John the little old.
Another interesting feature that suggests that [the little old woman] is an
integrated unit is that we can replace it by one word representing the whole
group and that we can use it as the answer to a question about the identity of
the person John kissed:
(4) John kissed her.
(5) -'Who did John kiss?'
- 'The little old woman.1
The grouping of words together which share a function is often referred to as
constituency. Structure in language can be described in terms of con-
stituency: complex language units (like the sentence) consist of a number of
constituents which, in turn, may consist of lower-level constituents.
Language structure is thus multilayered or hierarchic.
Despite the strong tendency for proximity in language, this principle may
be overridden by other considerations. Compare the following two sentences:
(6a) Sarah is painting her house.
(6b) Is Sarah painting her house?
Example (6a) expresses an activity in progress (is painting) enacted by
someone (Sarah) and involving an object (her house). It thus seems
reasonable to divide the sentence into the following parts: [Sarah], [is
painting] and [her house]. That is and painting form a group seems
intuitively right. Nevertheless it is possible to move one of the words without
moving the other, as in example (6b), where is is moved up in front of Sarah.
The physical separation of the two words does not in any way impair the
sense that they belong together in a group, as a constituent. The 'broken
relationship' seems closely related to the communicative difference between
the two examples: the first sentence is a statement, the second is a question.
It thus seems that communicative function is a factor which may override the
principle of proximity. The term usually applied to a "broken relationship* in
language is discontinuity: in the second example, Is and painting form a
discontinuous group to serve a specific communicative purpose.
Syntax deals with the relationship between the units of a sentence, more
specifically the various constituency groupings (continuous as well as
discontinuous) that the units enter. Like morphology, syntax is part of our
intuitive linguistic knowledge.
2.6. Linguistic creativity and ambiguity

Our intuitive knowledge of syntax is not restricted to an ability to recognize
various word order patterns when we see them: we all know how to use them
whenever we engage actively in communication. Thus, as has been em-
phasized by proponents of a particularly influential school of grammatical
thought, Generative Grammar, we all possess the ability to understand and
produce new sentences, sentences which have never been uttered or written
before, simply by using the familiar patterns of syntax and the lexicon, i.e.
the words of the language. Some of the examples discussed in the preceding
sections are examples of this kind: not many native speakers of English are
likely to have come across them before. In this technical sense, language is
creative: although it contains a finite number of building blocks (the words
in the lexicon), the rules for their legitimate combination are such that an
infinite number of sentences can be produced. Maximal flexibility in
matching expression and meaning is thus ensured.
Another example of the open-endedness of language is the lack of iso-
morphism, i.e. the lack of a one-to-one relationship, between the units of
language and the items of the world that we discuss and refer to, using
language. One fairly trivial but instructive example of this is the fact that
most nouns can be used to refer to more than just one particular item in the
'real world': in appropriate contexts an expression like the car can be used
about any car, not just one car. Furthermore, car is so general in meaning
that it appropriately covers a fascinating range of past, present and future
vehicles. The units of language can be said to have a generic potential.
Yet another aspect of language, related to linguistic creativity, and which
involves syntax more directly, is the principled diversity of meaning we
sometimes encounter in a single expression. That ambiguity is indeed an
important factor in language becomes evident when we consider examples
like the following (taken from Bache 1985b: 56; Chomsky 1957: 88; Lyons
1968: 249; Schibsbye 1970: 30, Wells 1947: section 3; and others), which
have been the object of much attention in the linguistic debate:
Linguistic creativity and ambiguity 23
(1) Old men and women are invited to the party.

(2) Flying planes can be dangerous.
(3) She wants to marry a Norwegian who is rich.
(4) He left his wife to deal with the creditors.
(5) The girl found a book on Main Street.
In some of these sentences, the ambiguity is more obvious than in others. But
most of us will eventually recognize the different meanings potentially
expressed in these examples.
In Old men and women are invited to the party, the expression [Old men
and women] refers either to a group of old men and old women or to a group
of old men and of women of any age (young and old alike), depending on
whether we interpret the adjective old as a modifier of men and women or of
men alone.
In Flying planes can be dangerous, [Flying planes] is either a word-like
nominal expression for aeroplanes with primary stress on the first word (like
police force) or it is a clause-like expression with primary stress on the
second word referring to instances of the activity of flying a plane. The
ambiguity arises because the normal concord rules are neutralized in can:
when Flying planes is a word-like nominal expression it takes the plural (as
in e.g. Flying planes are dangerous); when it is a clause-like construction on
a par with to fly a plane, it takes the singular (as in e.g. Flying planes is
dangerous).
The example She wants to marry a Norwegian who is rich shows that there
are sometimes different interpretations of referring expressions: either [a
Norwegian who is rich] refers to a particular person (e.g. Knut Flo from
Oslo) or it refers to anyone who qualifies as a rich Norwegian, i.e. any
member of the class of rich Norwegians.
In He left his wife to deal with the creditors, the person referred to by He
either lets his wife deal with the creditors (i.e. his wife is the agent of to deal)
or he leaves his wife with the purpose of dealing with the creditors himself
(i.e. He is the agent not only of left but also of to deal).
Finally, in The girl found a book on Main Street, the girl either found a
book about Main Street, or it was on Main Street, of all places, that she
found a book. Either on Main Street is part of a more complex construction a
book on Main Street, in which it describes the subject matter of the book
involved, or it is a more independent construction describing the location
where the girl found the book.
Characteristically, as we have seen, the different interpretations of all the
examples described above are tied to different syntactic patterns (i.e.
alternative relationships between the units involved) or different uses of the
units making up the sentence. The recognition of ambiguity in such cases is

thus a sign that we have a fairly advanced, if Only' intuitive, knowledge of
syntax and grammar.
2.7. Competence and performance

In the preceding sections we have established the fact that the speakers of a
language have a high degree of linguistic sensitivity and informal knowledge
of their own language. In other words, they have what is often referred to as
linguistic competence. Not only are they capable of identifying grammatical
units like words and sentences, they also recognize complex syntactic
patterns and attach appropriate meanings to them, as witnessed in cases of
ambiguity. Most important of all, they know how to put their intuitive
knowledge to use whenever they engage in communication. More technically
speaking, they know how to turn their linguistic competence into actual
linguistic performance. The fact that few speakers of a language are capable
of describing their language skills and of defining the relevant units and
patterns of language in a rigid, principled manner should not make us
underestimate their competence. Language is in this respect similar to
activities like walking or riding a bike: we are very competent at doing these
things without thinking about how we do them. And most of us would be
hard put to describe all the exact movements involved in these activities in a
principled, scientific manner.
The intuitive knowledge speakers have of a language comprises much
more than a knowledge of its formal properties, i.e. linguistic competence.
Intuitively, we know not only how linguistic expressions are structured but
also how to use them appropriately in different contexts or situations and in
relation to our communicative intentions. For example, speakers of English
know how to be formal or informal in their verbal interaction with other
speakers. They also know how to describe events as located in time, how to
elicit information, how to refer to things and persons, and so on. In short,
they have a knowledge of how to do things with linguistic structures. The
overall intuitive knowledge that speakers have of a language and of how to
use it in context is called their communicative competence. Though com-
municative competence is largely to do with language in use, it is still
possible to regard it in terms of 'knowledge of language' and keep it clearly
apart from performance; for there is obviously a difference between what
speakers are capable of doing verbally and what they actually do in a given
situation (see Dik 1989: 5).
Once we have accounted for the structural properties of a well-formed
sentence, we have to consider in what context it is appropriately uttered.
Syntagmatic and paradigmatic relations 25
When describing a language we are therefore concerned not only with

syntactically and semantically well-formed sentences but also with the
appropriateness of sentences in a given context. For example, the near-
equivalent sentences You must make your payment by May 31st and Your
payment must be made by May 31st are both syntactically and semantically
well-formed, but in some contexts only the passive sentence is appropriate,
in others only the active. To account for a native speaker's choice of one
rather than the other, we need the concept of communicative competence,
which combines linguistic competence with context.
2.8. Syntagmatic and paradigmatic relations

In this section, we shall draw attention to a specific example of the kind of
knowledge that native speakers of a language seem to have, an important
aspect of their linguistic and communicative competence, the knowledge of
paradigms or choice relations.
As we have already pointed out, language is of necessity linear but has
syntactic structure imposed on it which exploits and overrides the linearity.
Consider a sentence like the following:
(1) Sally teaches grammar.
This sentence has a relatively simple syntactic structure involving the
horizontal relationship between the constituents [Sally], [teaches] and
[grammar]. According to the rules of English grammar, we interpret the
sentence as a statement to the effect that Sally is the one who teaches and
grammar is the subject taught. From a cross-linguistic, universal point of
view, there is no necessary single arrangement of constituents to express this
particular piece of information. In other languages, it may well have to be
expressed through a different arrangement of the constituents making up the
sentence, corresponding to, for example, Teaches Sally grammar or Sally
grammar teaches, which are ungrammatical in English. The kind of
horizontal relationship that can be established between the constituents of a
sentence is often referred to as Syntagmatic.
Each of the constituents in the Syntagmatic relationship in Sally teaches
grammar might have been more complex, thus adding to the overall
complexity of the sentence:
(2) The young woman is teaching English grammar.
To go from the first sentence to the second we replace [Sally] by [The young
woman], [teaches] by [is teaching] and [grammar] by [English grammar].
The basic structure of the two sentences is the same. Further complexity is of
course possible:
(3) The very beautiful young black American woman that you met at the pub last
night could have been teaching advanced English grammar.
In this sentence, there are again three main constituents corresponding to
those in the two first examples: [The very beautiful young black American
woman that you met at the pub last night], [could have been teaching] and
[advanced English grammar]. This means that despite the verbosity of this
example, its basic structure is like that of Sally teaches grammar: there are
three main constituents only.
The structural similarity of the three examples discussed above shows that
although language is linear, thus calling for the syntagmatic, horizontal
arrangement of the constituents in the sequence that we recognize as a
syntactic structure, there is at the same time a vertical dimension to language.
A sentence is not just a sequence of elements or units which enter some sort
of horizontal relationship. Rather, a sentence contains a number of slots
which may be filled in different ways for different communicative purposes.
Thus, at one level, the three examples contain the same number of slots,
namely three, but these slots are filled with constructions of different length
and complexity. The constructions which are possible in a particular slot
(e.g. [teaches], [is teaching] and [could have been teaching]) enter a choice
relation: they are all candidates for a particular function at a particular point,
and the choice of one excludes the others. The relationship between the
possible constructions in a particular slot is often referred to as paradigmatic.
The implication of all this is that the linearity of language should be
viewed in terms of a sequence of slots, each an important hallmark, at which
the language user has a choice of expression. Language is both syntagmatic
and paradigmatic.
Sometimes the choice of expression for a given slot is a choice of one
lexical item rather than another:
(1) Sally teaches grammar.
(1') Sally teaches physics.
The paradigmatic choice between grammar and physics in the frame [Sally
teaches ] is a purely lexical choice, with no implication for the other
constituents of the sentence, and therefore not terribly interesting from a
grammatical point of view. Other paradigmatic choices involve grammar:
(la) Sally teaches grammar.
(1 b) Sally is teaching grammar.
(1 c) Sally taught grammar,
etc.
Syntagmatic and paradigmatic relations 27
The choice of verb form in a frame like [Sally grammar], where a

number of different forms of the verb teach are possible, must be accounted
for. A grammar of English must provide answers to questions like 'Why is
teaches but not teach all right in that particular frame?', 'What is the
difference between teaches and taught? Or between teaches and is teach-
ingT etc. Such questions concern inflectional morphology and competing
syntactic constructions.
It is also important to specify in our grammar what types of construction
are possible in particular slots. As we have seen, instead of a name in the first
slot ([Sally]), we may have a group of words ([The young woman] and [The
very beautiful young .... last night], respectively), but we cannot normally
have a clause:
(4) *That Sally is very competent teaches grammar.
In other frames, clauses as well as names and groups of words are perfectly
possible in the initial slot of the sentence:
(5a) Sally surprised Jack.
(5b) The young woman surprised Jack.
(5c) That Sally is very competent surprised Jack.
It is important to realize that the two dimensions of language, the syntag-
matic and the paradigmatic, are closely interrelated. Thus the choice of a
particular construction to fill a particular slot may well affect later choices of
constructions (and, conversely, the choice of a construction may be made in
anticipation of choices one wants to make later on). Consider the following
pair of sentences:
(6a) The young woman teaches physics.
(6b) The young women teach physics.
Here the choice of the singular noun woman in the initial major constituent
necessitates the choice of teaches rather than teach as the second constituent.
By comparison, the choice of the plural noun women in the initial major
constituent leads the speaker to choose teach rather than teaches later on in
the sequence. The term grammar should be interpreted in a broad sense
covering both the paradigmatic and syntagmatic dimension.
In conclusion, the intuitive knowledge that native speakers have, their
competence, includes knowledge not only of the syntagmatic dimension of
language but also of the paradigmatic dimension. As with the other aspects
of native speaker intuition dealt with in the preceding sections, it is difficult
to describe one's knowledge of language in a precise, appropriate and ob-
jective manner. The aim of this grammar is to provide such a description of
English. What we have set out to do is not simply to teach you, our readers,
grammar, because in the sense discussed in the preceding sections you know
a lot of grammar already. What we want to do is rather to offer the tools, the
terminology and the insights necessary for making your knowledge more
explicit.
2.9. Recapitulation
In this introduction to syntax we have shown that speakers of a language
have a high degree of linguistic competence: they have an intuitive, implicit
knowledge of the basic units of grammar and the various relationships these
enter into. This linguistic competence includes intuitions about syntax (the
principles of linearity and proximity as well as the principle of constituency
which arises from and overrides the two other principles) and of grammat-
ically conditioned ambiguity. Part of the linguistic competence of language
users is also a knowledge of paradigmatic choice relations in language.
Despite this highly developed competence, most speakers are unable to
describe their language skills appropriately. They may be able to identify
words and sentences but they cannot define these units. They easily
recognize grammatical strings of words in contrast to ungrammatical ones,
possible paradigmatic choices in contrast to impossible ones, as well as
grammatically conditioned ambiguity. And above all, they know how to use
language appropriately: they have communicative competence. But again, if
prompted, most people would fail to offer an appropriate account of why and
how they do these things. Thus, when we speak of'learning the grammar of a
language1, it is not simply a question of acquiring new knowledge but also a
question of becoming more conscious of something that we know intuitively
already. Even to the foreign learner in need of getting 'all the facts of the
language' right, the process of learning grammar to some extent involves
getting intuitive linguistic and communicative knowledge turned into explicit
conscious knowledge.
In order to teach (native as well as non-native) speakers of English the
grammar of the language, we need to turn the intuitive linguistic and com-
municative competence that native speakers of English have into an explicit
one. Against this background, the aim of the present grammar is to offer an
appropriate descriptive apparatus and to present the relevant rules of
competence that native speakers of English employ when they engage in
linguistic performance, i.e. in actual communication.
3. Elementary sentence analysis
3.1. The basic form and function approach

As a first step towards establishing an appropriate descriptive apparatus for
the grammar, we shall introduce a fairly elementary approach to the
description of the constituents making up sentences. The specific aim of this
is to provide the grammar with a common framework and terminology.
3.1.1. Form and function

We draw a basic distinction between the form and the function of constitu-
ents, which applies to all levels of description. We have already touched on
this distinction. In our discussion of paradigmatic choice relations in section
2.8 above, we noted that different types of construction may fill a particular
slot in a sentence. Thus, as we saw, [Sally], [The young woman] and [That
Sally is very competent] are all possible choices in the empty slot in the
frame [ surprised Jack]. Another way of formulating this insight is to
say that different forms may assume the same function in a sentence: [Sally],
[The young woman] and [That Sally is very competent] are different forms
but may perform the same syntactic function relative to surprised Jack.
3.1.2. Sentence functions

The main slots for which there is a choice of form in a sentence frame are
called sentence functions. We recognize five basic sentence functions:
S = subject
P = predicator
Ο = object
C = complement
A = adverbial
We employ two different techniques in our structural representations: linear
analyses and tree diagrams. The two techniques are notational variants, i.e.
different ways of showing the same structure. In a linear analysis (which is
convenient for simple or partial analyses in run-on texts), we use square
brackets to indicate the beginning and the end of constituents, each bracket
tagged with the appropriate label of analysis. In a simple sentence like Sarah
laughed, we can identify two sentence functions: a subject (Sarah) and a
predicator (laughed). The linear analysis looks like this: ^[Sarah] p[laughed].
30 Elementary sentence analysis
The tree diagram, which is a conventional form of syntactic representation,

provides an accessible overview of complex analyses. Using the label 'Sent'
for sentence, we can draw the following tree diagram for Sarah laughed:
Sent
Sarah laughed
The lines slanting downwards from Sent indicate a 'consist-of relationship1:

Sent consists of S and P. We use triangles to indicate that our analysis of
Sarah and laughed is incomplete: we have not assigned the appropriate form
labels yet. Before we do so, let us first look at some more examples of the
five sentence functions:
(1) The old man wrote a long letter.
s
[The old man] p[wrote] °[a long letter]
Sent
The old man wrote a long letter
(2) He was writing very slowly.

S[He] p[was writing] A[very slowly]
Sent
S
A
He was writing very slowly
(3) The letter was unbearably long.

s
[The letter] p[was] c[unbearably long]
Sent
The letter was unbearably long

The basic form and function approach 31
3.1.3. Four form types

There are four different form types capable of assuming sentence functions:
w = single word
g = group of words
cu = compound unit
cl = clause
For example, in Sarah laughed, both the subject and the predicator are single
words: s[Sarah] p[laughed]. In The old man wrote a long letter, the subject
and the object are groups of words, the predicator a single word: s[The old
man] p[wrote] °[a long letter]. All four types of form are present in an
example like That Helen left the party so early had bothered Jack and Jill
immensely: the subject [That Helen left the party so early] is a clause; the
predicator [had bothered] is a group; the object [Jack and Jill] is a compound
unit with two elements linked together, or coordinated; and, finally, the
adverbial [immensely] is a single word.
Notice that we use lower case letters for forms, capital letters being
reserved for functions. Separating the two by a colon (:) we have a
convention for describing both the function and the form of a constituent:
S:cl [That Helen left the party so early]
P:g [had bothered]
O:cu [Jack and Jill]
A:w [immensely]
The colon convention is used mainly for simple or partial analyses in run-on
texts. For more complex structures displayed in tree-diagrams, we use a
function-over-form convention:
Sent
That Helen left had bothered Jack and Jill immensely

the party so early
The function-over-form convention is used to indicate that a sentence

constituent has a function 'upwards in the tree" in relation to the other
constituents of the sentence, while internally it is a construction of a certain
form type to be further analysed 'downwards in the tree' (unless of course it

is a single word, permitting no further syntactic analysis).
Here are some more examples:
(1) She had promised that they would come.
S:w
[She] P:g[had promised] O:cl[that they would come]
Sent
She had promised that they would come
(2) The farmers laughed and danced until morning arrived.

S:
8[The farmers] P:cu[laughed and danced] A:cl[until morning arrived]
Sent
The farmers laughed and danced until morning arrived
3.1.4. Word classes

One of the forms introduced in section 3.1.3 will now be specified further:
the individual word (w). Words are traditionally divided into eight main
word classes according to their notional and formal characteristics:
n = nouns (e.g. car, letter, Jack, idea)
v = verbs (e.g. write, be, receive, hear)
adj = adjectives (e.g. long, old, afraid, big)
adv = adverbs (e.g. slowly, gently, duly, very)
pro = pronouns (e.g. he, she, who, any, this)
prep = prepositions (e.g. by, at, to, from, in)
conj = conjunctions (e.g. that, because, although)
art = articles (the, a, an)
Nouns typically express things or persons. In doing so they are often

combined with articles and inflected for the expression of number (e.g. the
car vs. the cars) and the genitive case (e.g. Jack vs. Jack's).
Verbs typically express actions (e.g. 'writing') or states (e.g. "being") and
inflect for tense and aspect (e.g. write vs. wrote), person and number (e.g.
write vs. writes).
Adjectives typically express qualities in relation to (pro)nouns (e.g. a long
letter /Jack is old) and often allow comparison (e.g. longer, longest I more
afraid, most afraid).
Adverbs typically express qualities in relation to verbs (e.g. Jack moved
slowly), adjectives (e.g. very big), other adverbs (e.g. §o_ gently), or the rest of
the clause (e.g. Fortunately, everybody was saved). Adverbs are often
derived from adjectives by means of the suffix -ly: e.g. slow —» slowly, gentle
—> gently. Like many adjectives, many adverbs allow comparison (e.g. more
slowly, most slowly).
Pronouns are a rather heterogeneous word class, comprising personal pro-
nouns (/, me; you; he, him; she, her; it, etc.), possessive pronouns (my, mine;
your, yours, etc.), reflexive pronouns (myself, yourself, herself, etc.),
demonstrative pronouns (that, those, this, these), interrogative and relative
pronouns (e.g. who, which, what) and indefinite pronouns (some, something,
any, anybody, no, nothing, every, everyone, all, (n)either, both, etc.).
Prepositions express relations (often spatial relations) between consti-
tuents. They typically do so by relating a noun or group (e.g. the table) to
another noun or group (e.g. the book) as in the book on the table, or to some
action or state (The book was placed on the table I The book is on the table).
Conjunctions also express relations between constituents. They do so either
by combining constituents at the same level (e.g. cars and books, clever but
arrogant) or by placing one clause (e.g. He didn't support her) at a lower
level in relation to another clause (e.g. I said that he didn't support her).
Articles typically combine with nouns to express definiteness (e.g. the car,
the idea) or indefmiteness (e.g. a car, an idea).
To the eight main word classes we may add intj (interjections like huh,
ouch, well, oh, wow, etc.) and num (numerals like^zve, hundreds, 1993, tenth,
twenty-first, etc.). The infinitive marker to is special: like many adverbs it is
obviously related to verbs; like auxiliary verbs such as may, can, will, etc., it
is placed in front of verbs (and thus in fact also resembles the articles, which
always precede nouns); like the conjunction that it seems void of meaning;
and formally it looks like the preposition to. We treat it separately, as a word
in its own right, and use the abbreviation infm to mark it in our analyses.
Each word class will be dealt with more elaborately in later chapters. At
this point we shall merely point to certain important facts relating to the
division of words into classes: (i) the identification of word-class member-

ship; (ii) the distinction between words as lexical items and words in use; and
(iii) open word classes vs. closed word classes.
A) The identification of word-class membership. It is often difficult to
classify a word in isolation from its linguistic context. Many words are of
course easily identifiable as members of one, and only one, word class:
policeman is always a noun, eliminate is always a verb, the always an article,
always always an adverb, etc. But there are also cases where we have to rely
on the context to reveal the function of the word before we can classify it.
Put differently, there are cases where word-class membership cannot be
determined independently of function. For example, blow is a noun in It was
a hard blow to him, but a verb in The referee may blow his whistle any time
now. Early is an adjective in He took an early train but an adverb in He left
the party very early. Down is particularly versatile: it is an adverb in The ship
went down, a preposition in Sally was walking confidently down the street, an
adjective in He is in one of his down periods at the moment, a verb in He
could down a pint of beer in twelve seconds. It may even be used as a noun
in the plural, as in He has his ups and downs, or with a completely different
meaning, as in The pillow was full of soft down.
As can be seen, it is necessary to distinguish, on the one hand, between
completely different words with the same form, and, on the other, between
different uses of what basically appears to be the same word. Thus it would
be sensible to say that down with the meaning 'first, soft feathers of young
birds' and down with a directional meaning are two different but homonym-
ous words, i.e. different words which happen to have the same manifestation
form. But then, what about the different uses of down with directional
meaning mentioned above: are they to be considered separate words? It
seems most appropriate to recognize the various functional realizations of
directional down as word-class distinct items (adverb, preposition, adjective,
etc.). In practice, then, we treat them as distinct but very closely related
words.
B) The distinction between words as lexical items and words in use.
Consider now the problem posed by the following examples:
(1) We all love Sally.
(2) Richard probably loves her more than the rest of us.
(3) Even bad-tempered, old Graham loved her once.
(4) As for myself, I cannot help loving her, too.
What we see here is formal (inflectional) variation of an item which does not
result in a change of word class. Though formally distinct, love, loves, loved
The basic form and Junction approach 35
and loving 'belong' to the same word, or lexical item, the verb love. This
means that we have to distinguish between a word in isolation - the base
form as it appears in a dictionary - and its inflectional manifestation form in
actual speech or writing. Henceforth we shall use capital letters when we
want to emphasize the status of a word as a base form and italics when we
want to emphasize the status of a word as a realized manifestion form: love,
loves, loved and loving are manifestation forms of the base form LOVE. We
use this convention in connection with verbs, nouns, adjectives and adverbs.
C) Open word classes vs. closed word classes. Of the eight main word
classes listed above, the first four (nouns, verbs, adjectives and adverbs) are
open word classes whereas the last four (pronouns, prepositions, conjunc-
tions and articles) are closed classes. Numerals and interjections are open
classes, while the infinitive marker is unique and thus does not fit into the
open/closed distinction at all.
Open and closed word classes can be distinguished in several different
ways. While open word classes have indefinitely many members, closed
word classes have relatively few members. While open word classes have a
fairly relaxed 'membership policy', admitting new members whenever there
is a need for them, closed word classes rarely allow any change. Thus we
often get new nouns (for example, as the result of new technology: LASER,
VIDEO, SOFTWARE, etc.) but the classes of prepositions and articles stay the
same for a very long period of time.
Members of open word classes typically have one or more independently
identifiable meanings, and there is no necessary semantic relationship be-
tween the meaning of one member of a class and another member of the
same class. Thus, simply by looking at nouns like POLICEMAN and STORY
we get a clear sense of their meaning. At the same time there seems to be no
obvious semantic relationship between them. Members of open word classes
are used by the speaker to instruct the hearer to think of things, events,
qualities, etc. that the speaker wants to talk about. By contrast, members of
closed word classes seem to have little independent meaning: they are
grammatical function words, assuming their meaning in relation to other
words. For example, in isolation it makes little sense to discuss the meaning
of, say, the definite article the, the conjunction that, the relative pronoun
which and even the preposition at. In appropriate linguistic contexts, how-
ever, these words assist open-class words in forming coherent sentences and
utterances. The presence of e.g. the definite article in the context of a
singular noun typically ensures a reading of the noun as a word which refers
to a specific, identifiable entity. Unlike open-class words, closed-class words
often enter a tight network of functional interdependences and relationships.
Thus, the function of the definite article is largely complementary to that of

the indefinite article: together they share a functional domain. The same is
true of the other closed word classes, though of course there are more
members and therefore more complex networks and systems.
3.1.5. Simple complete analyses

We are now in a position to offer complete analyses of sentences consisting
of one-word constituents, like the following (note that in such cases we no
longer need the triangle convention):
(1) John left her yesterday.
S:n
[John] P:v[left] O:Pro[her] A:adv[yesterday]
Sent
S P Ο A
n V pro adv
I 1 1 1
bhn left her yesterday
(2) Predictably, everybody liked chocolate.

A:adv
[Predictably] S:Pro [everybody] P:v[liked] O:n[chocolate]
Sent
Predictably everybody liked chocolate
For sentences which contain complex constituents we still use the triangle
convention to indicate that further analysis is possible:
(3) Marion said it was just as well she had gone.
S:n
[Marion] P:v[said] O:cl[it was just as well she had gone]
Sent
Marion said it was just as well she had gone

3.1.6. Discontinuity
As pointed out in section 2.4, there is a strong tendency in language for
constituents which belong together to be positioned together. However, this
principle of proximity is violated under well-defined conditions (see section
5.6 below). In both our linear analyses and our tree diagrams, the resulting
discontinuity is marked by hyphens in the following way:
(1) Ildiko did not send the letter last night.
S:n
[Ildiko] P;g-[did] A:adv[not] -P;g[send] O:g[the letter] A;g[last night]
Sent
Ildiko did not send the letter last night
(2) Have they ever met Francis?

p;
g-[Have] S:Pro[they] A:adv[ever] -P:g[met] O:n[Francis]
Sent
A -P O
g- pro adv -g
Δ
Have they ever
Δ
met Francis
In these examples, right-hyphenation (i.e. hyphenation after a label, such as

P:g- in example (1)) indicates a discontinuous relationship between the unit it
represents in the tree (did) and a unit in the subsequent linguistic context
(send), identically labelled but with left-hyphenation (i.e. a hyphen before the
label, such as -P:g in example (1)).
Notice that only one hyphen is used for each part of the discontinuous
constituent in our linear analyses, representing both discontinuous form and
discontinuous function (e.g. T:g-' for Have in example (2)), whereas in our
tree diagrams two hyphens are used, one for the form label (e.g. 'g-') and one
for the function label (e.g. T-'). It is also important to notice that although
each part of a discontinuous constituent may consist only of one word, as in
all of the examples above, we have not yet reached word level in our
analysis. The internal relationship of the parts that have been separated
remains to be specified, exactly as in continuous constituents.
Having introduced the main sentence functions and the main types of form
manifesting them, as well as the convention for marking discontinuity, we
now turn to each of the functions S, P, O, A and C.
3.2. Sentence functions and sentence structures

3.2.1. The predicator
The identification of subject, object, complement and adverbial often
depends on the prior identification of the predicator. Fortunately the form of
the sentence predicator is relatively stable and therefore fairly easy to
identify. It always consists of one or more verbs:
(la) Jack treated Sophia very badly.
(1 b) Jack is treating Sophia very badly.
(Ic) Jack has been treating Sophia very badly.
(1 d) Jack may have been treating Sophia very badly.
(1 e) ?Sophia may have been being treated very badly by Jack.
As we see in these examples, there are various ways of expressing a situation
of'Jack treating Sophia very badly': the key word is in each case TREAT. In
fact, the italicized predicator in (la) to (le) can be regarded as different
manifestation forms of the base form TREAT, involving one or more words.
To describe the organization of the predicator, we distinguish between yü//
verbs and auxiliary verbs. A predicator may consist of just a full verb (as in
example (la)) or a full verb as head preceded by up to three (in exceptional
cases: four) dependent auxiliary verbs (as in examples (Ib) to (le)).
The difference between a full verb and an auxiliary is normally one of
semantic weight: full verbs have independently identifiable lexical meanings
whereas auxiliaries have functional characteristics like closed-class items
(articles, prepositions, pronouns and conjunctions), relating to and modifying
full verbs. Formally, the two types of verb can be distinguished in terms of
linear position: in predicators where both are present, the last verb is almost
always the full verb and the others are auxiliaries (we disregard cases of
inversion like Also killed in the shootout were three teenagers from the
Bronx, see section 5.3.7). If one wants to test whether a verb is a full verb or
an auxiliary, one can convert a statement containing the verb into a ^yes-no
question' (i.e. a question of the type which tries to elicit either a yes or a no
for an answer). If the verb readily precedes the subject it is an auxiliary
whereas if it cannot precede the subject it is a full verb:
Sentence functions and sentence structures 39
(2a) Rob was having a nightmare.

(2b) Was Rob having a nightmare?
(3a) He can run a mile in six minutes.
(3b) Can he ru/i a mile in six minutes?
(4a) Steven finished his cheeseburger.
(4b) ^Finished Steven his cheeseburger?
(5a) Cathy kept laughing.
(5b) *Kept Cathy laughing?
In examples (2) and (3) BE and CAN are shown to be auxiliaries. In (4) and
(5) FINISH and KEPT are shown to be full verbs. To form a yes-no question
from a statement containing a full verb we have to use DO-support:
(4c) Did Steven finish his cheeseburger?
(5c) Did Cathy keep laughing?
In such cases DO is an auxiliary.
Note that three verbs, BE, HAVE and DO, are special in that they function
sometimes as auxiliaries and sometimes as full verbs. In the latter case they
may stand alone in the predicator:
(6a) Jack is now fully awake.
(7a) The old dancer has fond memories of Paris.
(8a) Her parents did nothing to change her mind.
The three verbs form a small closed class of so-called primary verbs. When
functioning as a full verb, BE regularly precedes the subject in yes-no
questions; HAVE occasionally allows this position in formal BrE; DO always
takes DO-insertion:
(6b) Is Jack now fully awake?
(7b) Has the old dancer fond memories of Paris?
(7c) Does the old dancer have fond memories of Paris?
(8b) Did her parents do nothing to change her mind?
The following central modal verbs always function as auxiliaries: can, could,
may, might, shall, should, will, would, must. Note that they have no base
form, only a fixed present and past form. There can never be more than one
central modal auxiliary in a predicator. In strings of auxiliaries, the others are
typically forms of the primary verbs BE and HAVE.
A predicator isfmite if it contains a finite verb. A predicator is nonfmite if
all the verbs in it are nonfmite. A finite predicator may contain up to three
(occasionally four) nonfmite verbs in addition to the finite verb. The
distinction between finite and nonfmite hinges on the presence or absence of
present/past marking: a finite verb is either formally present or formally past

whereas a nonfinite verb belongs to one of the following three form types:
(1) infinitives (with or without the infinitive marker): (to) break, (to) think,
(to) -worry, etc.;
(ii) present participles (breaking, thinking, worrying, etc.);
(iii) past participles (broken, thought, worried, etc.).
In the following examples, all the predicators (marked in square brackets) are
finite, containing a finite verb (in italics):
(9) Jack and Jill [take] a walk every morning.
(10) Jack [takes] things as they come.
(11) Jack and Jill [have taken] their stand on the issue.
(12) Both of them [could have been taking] the book to the library.
All the non-italicized verbs in these finite predicators (i.e. taking, taken,
have, been) are nonfinite by themselves. The same is true of take when it is
an infinitive, not a present form, as in the following examples:
(13) To take a walk would be foolish.
In a string of verbs in a finite predicator, it is the first verb (the first
auxiliary) which is finite. This verb is often referred to as the operator. To
form yes-no questions there is subject-operator inversion, often referred to as
partial inversion because only a part of the predicator is moved, cf. examples
(2) and (3) above; for discussion see section 5.3 below.
Note finally that there can be only one full verb in a predicator. In
examples like the following the second full verb is thus by definition outside
the sentence predicator:
(14) My old friend [decided] to leave the party.
(15) His girlfriend [stopped] singing.
In these examples, to leave and singing are part of, or fully constitute, the
object rather than belong to the predicator (see section 4.3.4).
Let us summarize the defining characteristics of the sentence predicator:
(i) A sentence predicator is always finite, containing a finite verb, showing
formal present/past marking.
(ii) A predicator contains one, and only one, full verb. In a predicator group,
the full verb always assumes head function.
(iii) Apart from the full verb, a predicator may contain up to three
(occasionally four) dependent auxiliary verbs (a modal auxiliary and/or one
or more forms of the primary verbs BE, HAVE and DO).
3.2.2. The subject

Once the predicates of a sentence has been found, it is usually fairly simple
to locate also the subject. Typically, the subject expresses the person or thing
which the predicator says, or predicates, something about. The subject is thus
the topic of statements, whereas the predicator is part of what is being stated
about the subject, the comment made about the subject. We can find the
subject by asking 'Who or what' immediately followed by the predicator, i.e.
'Who or what P?' The answer to that question is the subject. Consider:
(1) The parish vibrated with gossip the next day.
(2) It was a terrible shock to Mummy and Daddy.
(3) Daphne had enjoyed the illicit character of our relationship.
To find the subject in (1) to (3) we simply ask the question 'Who or what P?':
(Γ) Who or what vibrated*? The parish (did)
(2') Who or what was? It (was)
(3') Who or what had enjoyed! Daphne (had)
While this fairly simple test applies to the vast majority of sentences, there
are instances where it does not really make sense to ask 'Who or what P?':
(4) It was raining cats and dogs.
(41) Who or what was raining? *Λ
More formally, the subject displays a number of defining characteristics:
(i) The subject typically precedes the predicator in simple statements (as we
see in examples (1) to (4)).
(ii) The subject is always placed between the operator and the rest of the
predicator in yes-no questions, immediately following the operator (if the
predicator is a primary verb it also immediately precedes the subject):
(la) Did the parish vibrate with gossip the next day?
(2a) Was it a terrible shock to Mummy and Daddy?
(3a) Had Daphne enjoyed the illicit character of our relationship?
(4a) Was it raining cats and dogs?
(iii) Like the predicator, but unlike any other constituent, the subject is
always obligatorily present in sentences expressing statements. This means
that minimal sentences expressing statements contain S and P only:
(5) John left.
(6) The last glimmer of hope evaporated.
(iv) There is concord between subject and predicator, i.e. agreement between
these constituents in terms of number ana person. With one exception (see
below), subject-predicator concord is restricted to the present form of the
finite verb: if the subject is in the singular third person (i.e. he, she, it, or
anything potentially represented by these pronouns), the verb takes the suffix
-(e)s, otherwise it appears in its base form:
(7a) I take it easy.
(7b) She takes it easy.
(8a) The young woman teaches English grammar.
(8b) The young women teach English grammar.
The verb BE is especially expressive with respect to concord, being the only
verb showing concord in the past form and showing three person distinctions
in the present form:
(9a) I am better now than I was.
(9b) You/We/They are better now than you/we/they were.
(9c) He/She/It is better now than he/she/it was.
(10) The book/books was/were far too expensive.
(v) With pronouns to which the distinction between the subjective and
objective case applies (e.g. I/me, he/him, she/her, we/us, they/them), the
subjective case is used when the pronoun functions as the subject of a finite
predicator (see e.g. (7a-b) and (9a-c)).
(vi) Subjects, but not objects, complements or adverbials can be represented
by a pronoun in a so-called tag question:
(11) {Bob} gave them extra work, didn't hel
(12) {You and I} know better, don't we?
In some sentences there are two subjects, a provisional subject (Sp) and a real
subject (Sr). Only it and there may function as provisional subject:
(13) It was obvious that he disliked her.
Sp:pro[It] P:v[was] Cradj [obvious] Sr:cl[that he disliked her]
Sent
was obvious that he disliked her

(14) There were five books on the table.

Sp:pro[There] P:v[were] Sr:g(-five ^^ A:g[on me
Sent
There were five books on the table
In such examples, the provisional subject is semantically very light: it func-

tions as a grammatical prop word in subject position, merely representing the
real subject, which - for reasons to be discussed later - has been postponed.
In the case of it, it is usually possible to move the real subject back to subject
position, and in the case of there this is occasionally possible:
(15a) It was obvious that he disliked her.
(15b) That he disliked her was obvious.
(16a) There were five books on the table.
(16b) Five books were on the table.
Note that in sentences with it as the provisional subject, the real subject is
extraposed in the sense that it is placed at the end of the sentence, outside the
actual sentence structure, and can sometimes be deleted without this
affecting the grammaticality of the sentence (as in (15a): It was obvious). By
contrast, the real subject in sentences with there as provisional subject is
more closely integrated in the overall sentence structure: it is often followed
by other constituents and it cannot be deleted.
There is used as provisional subject in so-called existential sentences, i.e.
sentences expressing that something or someone exists somewhere. Existen-
tial sentences are always intransitive (i.e. Object-less', cf. sections 3.2.3 and
3.2.5 below). The real subject is in such sentences typically indefinite (e.g. a
book, books, some books, no book, something, somebody, nothing) rather
than definite (e.g. the book, these books, John 's books, that one, Bill, my
parents, etc.). The predicator is usually realized by a form of BE, by a modal
verb + be, or by SEEM, APPEAR, HAPPEN or TEND followed by to be:
(17) There were several students in the library.
(18) There could be more than one kind of complexity.
(19) There seemed to be no one left to talk to.
Other verbs are possible in existential sentences:
(20) There remained a few unsolved problems.
(21) There emerged in him a peculiar sense of affection for her.
Such constructions are fairly formal and the verbs are always semantically
light, denoting either something stative (as in (20)) or the transition or arrival
of something (as in (21)).
Occasionally the real subject of an existential sentence is realized by a
definite construction to express that a known entity is an example or a
possible solution to a problem, cf. the following data (from Swan 1995: 591):
(22) 'Who could we ask?' 'Well, there's James, or Miranda, or Ann, or Sue,...'
(23) 'Where can he sleep?' 'Well, there's always the attic.1
// and there as provisional subjects should be distinguished from it and there
with referential meaning. It and there with referential meaning relate to an
entity (object, place, etc.) in the 'external world' or in the preceding text:
(24) It came towards me with a fierce snarl.
S: ro
P [It] P:v[came] A:8[towards me] A:8[with a fierce snarl]
(25) There I finally found the letter.
A:pro[There] S:prO[I] A:adv[fmally] P:v[found] O:g[the letter]
In example (24), // obviously refers to something, e.g. an animal, in the

(linguistic and/or extralinguistic) context. In (25), There refers to the place -
contextually familiar to the speaker and hearer - where the letter was found
(e.g. Ί opened the top drawer and there I finally found the letter").
Occasionally we find cases of it used as a non-referential grammatical prop
word, i.e. with little or no meaning, as the only subject in a sentence
(especially in expressions about weather conditions, time and distance):
(26) // was raining again.
S:pro[It] P:g[was raining] A:adv[again]
(27) It was getting late.
S:pro|-It] P:g[was getting] C:adJ[late]
(28) It is a long way to Fitzroy Crossing.
S:pro[it] P:V[JS] C:g[a iong way to Fitzroy Crossing]
We employ the abbreviation 'S' whenever there is only one subject, whether
it is a grammatical prop word or a full referring expression. The abbrevi-
ations 'Sp' and 'Sr' are used only when there are two subjects.
3.2.3. Basic sentence structures

There are not only different types of subject but also different types of object
and complement. These relate closely to the nature of the predicator. In the
following, we shall therefore offer a subclassification of predicators allowing
for the relevant distinctions pertaining to objects and complements.
Predicators can be divided into three main classes: intransitive, transitive and
copula. An intransitive predicator is a predicator which takes no object or
complement. Some intransitive predicators take an obligatory adverbial
and/or a number of optional adverbials:
(1) (Again) Richard was sleeping (heavily) (in the room next door).
A:adv
[Again] S:n[Richard] P:g[was sleeping] A:adv[heavily] A:g[in the room
next door]
(2) Jessica was in London.
S-n[Jessica] P:v[was] A:8[in London]
Disregarding the possible occurrence of optional adverbials, we can represent
the possible basic sentence structures in statements with intransitive predic-
ators in this way:
SP
SPA
A transitive predicator is a predicator which takes an object:
(3) Richard kissed Jessica.
S:n
[Richard] P:v[kissed] O:n[Jessica]
(4) The naughty boy teased his parents at all times.
S;
g[The naughty boy]P:v[teased] O:8[his parents] A:ß[at all times]
Transitive predicators are thus associated with the following basic structure:
SPO
A copula predicator is a predicator which takes a complement:
(5) Marion is such a nice person.
S:n
[Marion] P:v[is] C:8[such a nice person]
(6) They looked so unhappy when I met them in Paris.
S: ro
P [They] P:v[looked] C:8[so unhappy] A:cl[when I met them in Paris]
Copula predicators are thus associated with the following basic structure:
SPC
It is important to note that many verbs may serve in more than one of these
basic types of predicator. Consider the following examples:
(7a) James smoked an expensive cigar after dinner.
(7b) Richard never smoked.
(8a) Sally was reading the newspaper when I got back.
(8b) Marion was reading while Tom did the dishes.
(9a) She got pretty mad at me.

(9b) I got little reward for my efforts.
In examples (7a) and (8a), smoked and was reading are transitive, taking an
object (an expensive cigar and the newspaper, respectively). In the
corresponding b-examples, these predicators are intransitive. In example
(9a), got is a copula predicator followed by a complement (pretty mad at
me). In (9b), got is a transitive predicator taking an object (little reward).
3.2.4. More sentence structures

In addition to the basic sentence structures identified in section 3.2.3 above,
there are a number of more complex structures which need to be introduced
in this section. Consider, first of all, the following sentences containing two
objects, a direct object (Od) and an indirect object (Oi):
(1) John gave the little girl a new doll.
S:n
[John] P:v[gave] Oi;g[the little girl] ^[a new doll]
(2) I bought my wife a new fur coat.
Srpropj P:vrbought] Oi:ß[my wife] ^«[a new fur coat]
In our sentence analysis we use the abbreviation if there is only one object
present in a sentence, reserving Od and Oi for sentences with two objects.
Predicators taking one object only are called monotransitive, those taking
two objects are called ditransitive. In monotransitive constructions, the
object is typically, though not inevitably, a direct rather than an indirect
object. In ditransitive statements, the indirect object normally precedes the
direct object:
S P Oi Od
Sometimes we find combinations of an object and a complement, or an
object and an obligatory adverbial, following the predicator (we return to the
distinction between obligatory and optional adverbials in section 3.2.9).
Predicators in such constructions are called complex-transitive. In complex-
transitive constructions, a complement is called an object complement (Co),
because it refers back to the object rather than to the subject:
(3) We painted the wall yellow.
S: ro
P [We] P:v[painted] O;g[the wall] Co:a%ellow]
(4) Actually, the staff elected Miss Johnson dean.
A:adv
[Actually] S:g[the staff] P:v
[elected] O:i[Miss Johnson] Co:n[dean]
The kind of complement that refers back to the subject in S P C structures is

often referred to more specifically as subject complement (Cs) in termino-
logical contrast to object complement.
Here are some examples of complex-transitive combinations of an object
and an obligatory adverbial:
(5a) My father put the book on the shelf.
S;
g[My father] P:v[put] O:8[the book] A:8[on the shelf]
(5b) *My father put the book.
(6a) I slipped the key into the lock.
S:pr0[I] P:v[slipped] O:g[the key] A:g[into Λε lock]
(6b) *I slipped the key.
Complex-transitive predicators thus yield two different types of structure:
SPOCo
SPOA
These sentence structures are obviously related to the copula S P C structure
and the intransitive S P A structure, respectively.
Before looking more closely at how objects, complements and adverbials
are identified, let us summarize the typical sentence structures in statements:
SP (intransitive predicator)
SPA (intransitive predicator)
SPO (monotransitive predicator)
S P Cs (copula predicator)
S P Oi d (ditransitive predicator)
S P O Co (complex-transitive predicator)
SPOA (complex-transitive predicator)
For discussion of the position of optional adverbials, see sections 3.2.9 and
5.5.
3.2.5. The direct object

In this section we take a closer look at transitivity and the identification of
direct objects. The term 'transitive1 implies both 'relation between entities'
and 'direction'. The nature of the relation established in a transitive
construction is expressed by the predicator and involves the constituents
subject and object(s). In section 3.2.2, we saw that the subject is commonly
the topic of simple statements. To this pragmatic function (i.e. general
communicative function) we can add one of a number of more specific
semantic functions that constituents may have in relation to the action or
situation expressed by the predicator. Typically, the subject has the role of
agent, i.e. it is the participant performing the action expressed by the
predicator. By contrast, the direct object (which like the predicator is
considered to be part of the comment made about the subject) is typically the
participant affected by the action expressed by the predicator, i.e. thing or
person towards which/whom the action is directed. We get a sense of these
roles or meanings when we consider examples like the following:
(1) Harris moved the bike.
S:n
[Harris] P:v[moved] O:8[the bike]
(2) The little girl kissed the shaggy dog.
S;
g[The little girl] P:v[kissed] O;g[the shaggy dog]
(3) They hit Sally on the head.
S: ro
P [They] P:v[hit] O:n[Sally] A:8[on the head]
In these examples, the subjects are agents performing the actions of'moving1,
'kissing1 and 'hitting', respectively. The objects are affected by these actions:
the bike gets moved, the shaggy dog gets kissed and Sally gets hit. That both
subjects and direct objects may express many other semantic functions is
evident in examples like (4) to (6):
(4) This bottle contains cold tea.
S:
g[This bottle] P:v[contains] O:S[cold tea]
(5) I saw her very clearly.
S; ro
P [I] P:v[saw] 0:Pro[her] A:8[very clearly]
(6) Max has received detailed reports.
S;n[Max] P:6[has received] ^[detailed reports]
A more detailed description of the various meanings and roles attached to
subjects and direct objects can be found in section 7.3.
From the typical semantic function of affected in a transitive relation we
can derive a simple question test for identifying direct objects similar to the
one we devised in section 3.2.2 for identifying subjects: the direct object in a
sentence may serve as an appropriate answer to the question 'Who(m) or
what' followed by the relevant partially inverted S P construction. This test is
in fact general enough to apply to all types of direct object, irrespective of
the semantic function assigned to the object. Thus, applying it to examples
(1) to (6) above, we get the following appropriate answers:
(1') Who(m) or what did Harris move? The bike
(2') Who(m) or what did the little girl kiss? The shaggy dog
(3') Who(m) or what did they hit? Sally
(41) Who(m) or what does this bottle contain? Cold tea
(51) Who(m) or what did I see? Her

(61) Who(m) or what has Max received? Detailed reports
There are a number of formal characteristics of direct objects:
(i) The direct object usually follows immediately after the predicator in
monotransitive statements, as in examples (1) to (6).
(ii) The direct object in a monotransitive active construction may often,
though not always, serve as the subject in an intransitive passive construc-
tion. The active/passive distinction is a voice distinction relating to informa-
tion structure (i.e. the way information is presented) and the assignment of
focus to constituents. Almost all the examples offered so far have been
active. Passive counterparts to active sentences are formed by assigning
subject function to the direct (sometimes indirect) object of the active
sentence and by changing the active predicator into a passive one consisting
of a form of BE followed by the past participle of the full verb of the active
predicator. The original subject may be expressed by an adverbial fty-group:
(7a) Jack's secretary typed Bill's letters.
S:
S[Jack's secretary] P:v[typed] O:g[Bill's letters]
(7b) Bill's letters were typed by Jack's secretary.
S:g[Bill's letters] P:8[were typed] A:8[by Jack's secretary]
(8a) The terrorists blindfolded everybody.
(8b) Everybody was blindfolded (by the terrorists).
(iii) With pronouns to which the distinction between the subjective and
objective case is used when the pronoun functions as direct object:
(9) She remembered him for his good manners.
S: ro
P [she] P:v[remembered] O:Pro[him] A:8[for his good manners]
(10) They saw us from a mile off.
In some sentences there are two direct objects, a provisional direct object
(Op) and an extraposed real direct object (Or). Only it may serve as
provisional direct object:
(11) I take it that he will not be present.
Sent
S _P Op Or
pro v pro cl
take it that he will not be present

(12) They found it difficult to work with him.

S: ro
P [They] P:v[found] °P:Pro[it] Co:adJ[difficult] O^fto work with him]
The provisional direct object functions as a grammatical prop word in object
position (immediately after the predicator), representing the extraposed real
object. An extraposed real object is always a clause. It is not usually possible
to move the extraposed object to normal object position:
(13a) I made it clear to her that I accept no nonsense.
(13b) *I made that I accept no nonsense clear to her.
It as a provisional direct object should be distinguished from it as a real
direct object with referential meaning:
(14) They took it (= e.g. his loyalty) for granted.
S: ro
P [They] P:v[took] O:Pro[it] A:ß[for granted]
In this example, it refers back to something mentioned in the previous
discourse: They never questioned his loyalty. I think they took it for granted.
Occasionally, in more or less fixed expressions, we find it used as a non-
referential grammatical prop word, i.e. with little or no meaning, as the only
object in a sentence:
(15) They hit it off together.
S: ro
P [They] P:v[hit] O:Pro[it] A:adv[off] A:adv[together]
(16) I like it here.
Siproj-jj P:v[iike] Oiprorjt] A:adv[herej
In our sentence analysis we employ the abbreviation ' whenever there is

only one object (typically a direct object), whether it is a grammatical prop
word or a full referring expression. Op' requires the presence of Or', and
vice versa. Similarly Od' requires the presence of Oi', and vice versa.
3.2.6. The indirect object

An indirect object typically expresses the participant benefiting or suffering
from the situation or action expressed by the combination of subject,
predicator and direct object. The semantic function assigned to the indirect
object is called beneficiary:
(1) Fred bought Sally a bunch of roses.
S:n
[Fred] P:v[bought] Oi:n[Sally] Od;g[a bunch of roses]
(2) Her little sister dealt Jack a severe blow.
S;
g[Her little sister] P:v[dealt] Oi:n[Jack] Od:ß[a severe blow]
But sometimes the semantic function of indirect objects is less concrete:
(3) My wife gave going to France a good deal of thought.

S:
8[My wife] P:v[gave] Oi:cl[going to France] Od:8[a good deal of thought]
In addition to the semantic clue, there are a number of other characteristics of
indirect objects which help us identify them:
(i) The indirect object usually appears immediately after the predicator and
immediately before the direct object, as in (1) to (3). There is, especially in
BrE, an exception to this rule: when functioning as direct objects, it and them
are sometimes seen to precede pronominal indirect objects:
(4) I gave it him.
S:pro[i] P:v[gave] Od:pro[it] Oi:pro[him]
A third option is often chosen here to avoid ambiguity or confusion: an

expression containing an adverbial -phrase instead of an indirect object, as
in I gave it to him.
(ii) Related to this last point is the general paraphrase relation between
ditransitive constructions and monotransitive (or complex transitive)
constructions containing an adverbial to- oryör-phrase:
(5a) They offered the old man a new job.
S: ro
P [They] P:v[offered] Oi:8[the old man] od:8[a new job]
(5b) They offered a new job to the old man.
S:pro[They] P:v[offered] Od:g[a new job] A:8[to the old man]
(6a) My mother has baked us a chocolate cake.
(6b) My mother has baked a chocolate cake for us.
(iii) Like the direct object, the indirect object may often serve as the subject
in a corresponding passive construction. Thus, to a ditransitive construction,
we often find two alternative corresponding passive constructions, one -
much the commoner - with the active indirect object as the passive subject
and one with the active direct object as the passive subject:
(7a) Jack gave Sally a silver ring.
S:n
[Jack] P:v[gave] Oi:n[Sally] Od;g[a silver ring]
(7b) Sally was given a silver ring (by Jack).
S:n
[Sally] P:ß[was given] O:g[a silver ring] (A:8[by Jack])
(7c) A silver ring was given Sally (by Jack).
S:
8[A silver ring] P;g[was given] O:n[Sally] (A:8[by Jack])
When the active direct object becomes the passive subject, there is a
preference for expressing the beneficiary in an adverbial to- oryör-phrase:
(7d) A silver ring was given to Sally (by Jack).
(iv) With pronouns to which the distinction between the subjective and
objective case is used when the pronoun functions as indirect object:
(8) Granny was reading them a chapter.
(9) Mr Smith ordered me a new radio.
Note finally that from ditransitive constructions we can typically only derive
monotransitive constructions with the direct object:
(1 Oa) Billy gave his brother a penknife.
(1 Ob) Billy gave a penknife.
(lOc) *Billy gave his brother.
With a few verbs, however, either the direct object or the indirect object in a
transitive construction may function as the sole object in a monotransitive
construction, cf. the following examples with TEACH:
(11 a) The young man taught us linguistics,
(lib) The young man taught linguistics,
(lie) The young man taught us.
Note that in clauses with TELL or TEACH, clausal direct objects require the
presence of an indirect object:
(12a) The young man taught us how to fix a lock.
(12b) *The young man taught how to fix a lock.
(13a) Jack told us that he missed his brother.
(13b) * Jack told that he missed his brother.
In our sentence analysis we only use the labels Od and Oi in ditransitive
constructions. In monotransitive constructions, we always use O, irrespective
of the potential function of the constituent in ditransitive constructions.
3.2.7. The subject complement

A subject complement (for which we use the abbreviation 'Cs' or simply "C")
expresses further information about the referent of the form realizing the
subject of the sentence:
(1) This is a misfortune.
S: ro
P [This] P:v[is] C;g[a misfortune]
(2) Jack became my best friend.
S:n
[Jack] P:v[became] C:8[my best friend]
(3) My brother looks very intelligent.
S:
8[My brother] P:v[looks] C:g[very intelligent]
Sentence junctions and sentence structures 53
(4) My sister often gets more upset than Jack.

S:
8[My sister] A:adv[often] P:v[gets] C:g[more upset than Jack]
In examples (1) and (2), there is identity between the subject and the
complement, the complement offering a (different) description of the thing or
person referred to by the subject This orJack, respectively. In examples (3)
and (4), the complement simply assigns a quality to the subject, that of being
Very intelligent' and 'more upset than Jack', respectively.
One diagnostic feature of the complement function is that - unlike
subjects, objects and adverbials - it can almost always be realized by an
adjective or adjectival construction (such as very intelligent and more upset
than Jack in examples (3) and (4), respectively). Like subjects and objects,
complements are however often realized by nominal constructions. But when
a complement is realized by a nominal construction (as in examples (1) and
(2)) one way of deciding that it is indeed a complement rather than, say, an
object is to see if it can be replaced by an adjectival construction without this
affecting the acceptability of the sentence:
(Γ) This is unfortunate.
(2') Jack became very friendly.
Exceptions to this diagnostic feature are rare:
(5a) Stephen made (= became) a fine soldier.
(5b) *Stephen made very brave.
The verbs serving as copula predicators fall into two groups: the stative 'BE'
family and the dynamic 'BECOME' family. The members of the 'BE' family
express 'identity' or '(possible) current possession of quality or characteristic
feature' (e.g. BE, REMAIN, APPEAR, FEEL, LOOK, PROVE, SEEM, SOUND,
STAY, TASTE). Here are some more examples:
(6) She remained silent.
(7) The decision appeared all wrong to me.
(8) The directors simply proved far too inefficient.
(9) That sounds good to me.
(10) The cheese tasted sour.
The members of the 'BECOME' family express change and are resultative in
meaning (e.g. BECOME, GET, FALL, TURN, GO, GROW, RUN and SPRING,
particularly in set phrases). Here are some more examples:
(11) Within a week my mother fell seriously ill.
(12) Our teacher eventually went raving mad.
(13) I grew quite fond of her despite our differences.
(14) The river was running dry.

(15) The lock sprang open.
The descriptive meaning associated with S P C constructions can sometimes
also be expressed by intransitive constructions with adverbials; compare:
(16) She was happy.
S: ro
P [She] P:v[was] C:adJ[happy]
(17) She was in high spirits.
S: ro
P [She] P:v[was] A:8[in high spirits]
The approach adopted in such cases is to distinguish between adverbial and
complement on the basis of form: when the descriptive attribute is realized
by a preposition group (like in high spirits), it is an adverbial; when it is
realized by an adjective (like happy), it is a complement. This approach is
consistent with the way we distinguish between adverbial and indirect object
in sentences involving a beneficiary like / gave the book to John vs. / gave
John the book: in the former example, to John is an adverbial, hi the latter,
the noun John is an indirect object.
Note finally that many verbs functioning as copula predicators may serve
also as transitive and/or intransitive predicators:
(18) He proved the point by singing the ballad himself.
(19) My mother fell and hurt her knee badly.
(20) We grow oranges in our garden.
3.2.8. The object complement

Object complements (for which we use the abbreviation 'Co') express further
information about the referent of the object of the sentence:
(1) We painted the wall white.
S: ro
P [We] P:v[painted] O:S[the wall] Co:adJ[white]
(2) They drove Stephen mad.
(3) Jack considered Jane his closest friend.
In these examples there is a close relationship between the object and the
object complement very similar to the one between subject and subject
complement in S P C constructions: in (1) the wall becomes white; in (2)
Stephen gets mad; and in (3) Jane is Jack's closest friend. In other words,
there is in each case an implied S P Cs construction. Interestingly S P O Co
constructions are actually turned into S P Cs constructions in the passive
voice. Thus corresponding to (1) to (3) we get the following:
(1') The wall was painted white.

S:
S[The wall] P:8[was painted] C:adJ[white]
(2') Stephen was driven mad.
(3') Jane was considered his closest friend.
Note that object complements, like subject complements, are realized by
nominal constructions or adjectival constructions. As with subject comple-
ments, the option of being realized by an adjectival construction is a fairly
reliable diagnostic feature; but there are striking exceptions:
(4) His fellow students elected him president / *dutifiil.
(5) The Vice-Chancellor appointed Bill dean of the humanities / *responsible.
The stative/dynamic distinction applies also to the relation between object
and object complement. Thus in examples (1), (2), (4) and (5) above, the
relation is dynamic whereas in (3) the relation is stative. Here are some more
examples of a stative relation between object and object complement:
(6) The directors found him qualified.
(7) She always called me uncle.
The kind of relation which exists between object and object complement may
be expressed by other types of construction, notably S P O A constructions:
(8) Many people regard Jack as an eccentric.
S;g[Many people] P:v[regard] ^:n[Jack] A:g[as an eccentric]
(9) The professor took hervor a native speaker of English.
3.2.9. Adverbials
Traditionally, the adverbial is regarded as the default sentence function in the
sense that it characterizes any function at sentence level that is not a subject,
predicator, object or complement. Here are some examples:
(1) The Ford went into the East Sector just after midnight.
S:
g[The Ford] P:v[went] A;g[into the East Sector] A:ß(just after midnight]
(2) Unfortunately, his leg was broken in three places.
A:adv
[Unfortunately] S:8[his leg] P:8[was broken] A;g[in three places]
(3) As casually as she could, she told me about it.
A;
g[As casually as she could] S:Pro[she] P:v[told] O:Pro[me] A;g[about it]
(4) Really, he never even met this woman, because he never got out of prison.
A:adv
[Really] S:Pro[he] A:adv[never] A:adv[even] P:v[met] O:g[this woman]
A:c
l[because he never got out of prison]
By looking at examples like these we get a sense of the price we pay for
treating the adverbial as a default function: it is far more heterogeneous in its
range of roles than the other sentence functions. It is in fact so heterogeneous
that it is difficult, if not impossible, to define it positively with any degree of
precision. There are, however, a number of noteworthy general features
characterizing adverbials: optionality, mobility, multiplicity, functional and
semantic diversity.
Though adverbials participate as obligatory functions in the intransitive S P
A structure type (e.g. Jack was in London) and the complex-transitive S P O
A structure type (e.g. Sally put the book on the table), they are often far more
peripheral to sentence structure than the other functions: they typically occur
as syntactically optional constituents. Thus most of the adverbials in sen-
tences (1) to (4) above could in fact be left out without this affecting either
basic sentence structure or acceptability:
(1') The Ford went into the East Sector.
(2') His leg was broken.
(3*) She told me.
(4') He met this woman.
Syntactic optionality should not be mistaken for semantic optionality: when
we leave out syntactically optional adverbials, the meaning of the sentence is
often radically changed. Thus in (41) the basic meaning of (4) is completely
reversed. In the other examples, the adverbials offer additional information
without which the sentences 'merely' lose specificity.
That adverbials are often mobile (in the sense that they may freely appear
in more than one position in a sentence) can be ascertained in an example
like the following:
(5a) One night the couple returned from a party in a gay mood.
(5b) The couple one night returned from a party in a gay mood.
(5c) The couple returned one night from a party in a gay mood.
(5d) The couple returned from a party one night in a gay mood.
(5e) The couple returned from a party in a gay mood one night.
Basically these variant sentences mean the same thing but differ slightly in
terms of the focus of each constituent, i.e. in terms of the speaker's presenta-
tion of the information. However, we do not want to imply that anything
goes with respect to the position of adverbials. There are restrictions:
(6a) With diligence she has completed the play.
(6b) *She with diligence has completed the play.
(6c) ?She has with diligence completed the play.
(6d) ?She has completed with diligence the play.

(6e) She has completed the play with diligence.
Sometimes the change of position has obvious semantic repercussions:
(7a) Clearly Bill saw Jane.
(7b) Bill saw Jane clearly.
(8a) Quite frankly, Jack told me about all this.
(8b) Jack told me about all this quite frankly.
In example (7a) what is clear is the fact that Bill saw Jane (i.e. it was clear to
the speaker that Bill saw Jane), whereas in (7b) what is clear is Bill's visual
experience of Jane. In (8a), Quite frankly is the speaker's comment on the
rest of the sentence (Ί am telling you this quite frankly'), whereas in (8b) this
adverbial describes the manner in which Jack told me about all this. In other
words, in the a-examples the adverbials are speaker-oriented and thus strictly
outside the scope of the message conveyed, whereas in the b-examples they
are fully integrated in the message, modifying the meaning of the predicator
in terms of the manner in which the situation referred to is carried out.
With optional adverbials (unlike other functions), we get an impressive
multiplicity of occurrence within the sentence. Instead of simply saying:
(9a) Jack left the room,
we might say:
(9b) 1 Well, ^to tell the truth, ^last Monday, ^without really meaning any harm,
Jack *curiously enough "once again 'quietly left the room °for a few
minutes, "} without locking the door, ™in order to catch a glimpse of her.
Exactly how many optional adverbials we can get in a sentence seems more a
question of stylistic consideration than of grammatical principle, but even in
elegant speech or writing it is not unusual to have several.
This multiplicity of occurrence is clearly related to the functional and
semantic diversity of adverbials. It is impossible to give a complete survey of
the many uses and meanings of adverbials, but again there are certain
noteworthy characteristics.
Adverbials serve at least three main communicatively significant subfunc-
tions, traditionally referred to as adjunct, disjunct and conjunct. An adjunct is
fairly closely integrated in the sentence structure, typically relating closely to
the predicator and somehow modifying or specifying its meaning. Adjuncts
are commonly used to express 'negation', 'time', 'place' (including 'source' and
'direction'), 'manner', 'instrument', 'reason', 'purpose', 'condition', 'degree',
etc., as in the following examples:
(10) Owen did not believe in an afterlife.
(11) I intend to leave for Rome tomorrow.

(12) In the distance he heard the screeching tyres of a car.
(13) Ursula chuckled hoarsely at her own choice of expletive.
(14) He opened the lock with a small pen-knife.
(15) Walter left the party because he was angry with Enid.
(16) He rushed after her to explain what had happened.
(17) I will do it if you will.
(18) Ißilly agree that we ought to get rid of the vice dean.
As is evident in these examples, an adverbial serving as an adjunct is clearly
a sentence function on a par with the subject, the predicator, the object etc.,
despite its frequent syntactic dispensability. By contrast, an adverbial serving
as a disjunct takes on a broader scope in that it relates to the rest of the
sentence and not just narrowly to the predicator. Disjuncts are commonly
used to convey the speaker's or writer's comment on the information ex-
pressed by the rest of the sentence or on the style or form of the expression
itself, as in the following examples:
(19) James is undoubtedly a talented piano player.
(20) Fortunately, Iris was not swept up in the Women's Liberation movement.
(21) She stupidly tried to steer me off the subject of money.
(22) Henry went to prison, believe it or not.
(23) Strictly speaking, she should have reported the incident.
(24) Quite honestly, I do not want you to leave Hawaii yet.
As we see, disjuncts concern the external relationship between the speaker or
writer and the sentence and are thus more peripheral to sentence structure
than adjuncts, which concern the internal relationship between sentence
functions. Conjuncts are like disjuncts in being peripheral to sentence
structure but differ from them in expressing a relationship between the
sentence and its linguistic or non-linguistic context (thus often assuming a
conjunction-like function). Typically, conjuncts serve to relate the sentence
in which they occur to a previous sentence or to knowledge or experience
shared by the speaker and listener, or they are used as discourse initiators:
(25) He was beginning to feel better disposed towards Margaret. All the same, he
would never allow John near her.
(26) The rule seems to apply convincingly to all the Type A cases reviewed so
far. However, there are important exceptions in the Type B material.
(27) My wife is very busy this evening. Nevertheless she is likely to go to bed
very early.
(28) So how are you these days, Sally?
(29) Well, it feels good to be outside Makai Manor for a change.
The internal structure of complex forms 59
(30) In the first place, I never wanted to get involved. Secondly; I do not like
being bossed around.
Like the other sentence functions, adverbials will be dealt with more thor-
oughly at a later stage (see especially section 5.5).
3.3. The internal structure of complex forms

3.3.1. The group
As we have seen, sentence functions may be realized by four different form
types: a word, a group, a compound unit or a clause. Of these, only the word
requires no further syntactic treatment. The other three forms are syntact-
ically complex and must therefore be examined with respect to their internal
structure. In this and the following sections we shall deal with the internal
structure of complex forms.
A group always consists of a head (abbreviated as H) and one or more
dependents (abbreviated as DEP):
(1) may have been held
DEP
[may] DEP[have] DEP[been] H[held]
(2) an extravagant party
DEP
[an] DEP[extravagant] H[party]
(3) interestingly enough
H
[interestingly] DEP[enough]
(4) in her honour
H
[in]DEP[her honour]
Groups are subclassified according to the form realizing the head. Here is a
list of the most common kinds of group with a single word as head:
Verb groups have a verb as head:
(5) might have been dancing
DEP
[might] DEP[have] DEP[been] H[dancing]
Noun groups have a noun as head:
(6) the sad result of the affair
DEP
[the] DEP[sad] H[result] DEP[of the affair]
Pronoun groups have a pronoun as head:
(7) everyone I know
H
[everyone] DEP[I know]
Adjective groups have an adjective as head:

(8) extremely miserable
DEP
[extremely] H[miserable]
Adverb groups have an adverb as head:
(9) very beautifully
DEP
[very] H[beautifully]
Preposition groups have a preposition as head:
(10) to a small town
H
[to] DEP[a small town]
The head of a group is normally an obligatory characterizing element
without which the group would have been some other kind of group, or
simply ungrammatical. Dependents are obligatory or optional noncharacter-
izing elements. In principle, both heads and dependents may be realized by
all four types of form (the word, the group, the compound unit, the clause),
but there are many restrictions applying to the individual kinds of group.
Note that in our notational system we do not mark groups according to type
(e.g. 'ng' for noun group): such information is redundant because it can be
read off directly from the form of the head at the next level below. Examples:
S:
(11) 8[the pilot] (left the party early)
DEP:art[the] H:n[pilot]
J3
g
PEP
art n
ι
the pilot
ι
(12) (She is meeting) O:8[someone you do not know]
H: ro
P [someone] DEP:cl[you do not know]
g
H DEP
pro cl
someone you do not know

(13) (We had) O:8[quite a party]

DEP:adv[quite] H:g[a party]
quite a party
(14) S:S[His telling his wife about it] (does not help us much)
DEP:pro[His] H:cl[telling his wife about it]
PEP H
pro IT
His telling his wife about it
In examples (11) and (12) we have a noun group and a pronoun group, re-
spectively. In (13) we have a 'group group' (i.e. a group with a group as head)
and in (14) we have a 'clause group' (i.e. a group with a clause as head).
There is no simple head-dependent relationship characterizing all the dif-
ferent groups that we have examined. By marking heads and dependents we
simply indicate that there is some relationship rather than a particular kind of
relationship, the only common factor being the obligatory, characterizing
nature of the head. Compare:
DEP:ad
(15a) J[clever] H:n[girls]
DEP:art
(15b) [the] H:n[girls]
In (15a) the dependent adjective clever 'modifies' the head girls by attributing
a quality to the indefinite plural referent of the noun. In (15b) the dependent
definite article the 'determines' the head, i.e. helps establish the referent of
girls as a definite specific group of girls. Thus, the head-dependent relation-
ship in the two examples is very different. Consider also examples like:
DEP:v
(16) [may] H:v[leave]
H: ro
(17) P [anything] DEP:adJ[British]
H: re
(18) P P[to] DEP;g[his girlfriend]
In example (16) the head-dependent relationship is not unlike that in (15a) in

that the dependent modal verb may seems to modify the head verb leave. In
both cases, the head seems semantically more independent and thus weight-
ier than the dependent. By contrast, in example (17) the pronominal head
anything seems semantically somewhat reduced, giving way to the meaning
of the dependent adjective British. Finally, in example (18), it makes little
sense to speak of either modification or a difference of semantic weight.
Rather, the head preposition to expresses the nature of the relationship of the
dependent his girlfriend to constituents outside the group.
It is necessary at this point also to comment briefly on the distinction
between groups and single words. With the exception of preposition groups,
groups can be viewed as expansions of single words (cf. e.g. Halliday 1994:
179ff). For example, the noun group the water is an expansion of the single
noun water, the adjective group very honest is an expansion of the single
adjective honest, and so forth. Often single words and groups are used in the
same way (e.g. with the same syntactic function, as in She was honest /Her
sister was very honest) and display very similar grammatical features (e.g.
cars and the cars, which are both plural expressions). Another way of
formulating this fact is to say that single verbs, nouns, pronouns, adjectives
and adverbs exhibit a strong potential for being the head of a group. Given
this close relationship between single words and groups, it is often con-
venient to be able to refer to both form types with a single label. In
traditional grammar there was a strong word-orientation: grammarians talked
about e.g. nouns, often implying also noun groups. In modern linguistics,
there is a tendency to focus more on the group level: grammarians now often
talk about e.g. noun groups (or 'noun phrases'), implying also single nouns.
In addition to these practices, which are difficult to avoid completely, we
shall operate with the following explicit cover terms:
verbal = verb group or single verb
nominal = noun group or single noun
pronominal = pronoun group or single pronoun
adjectival = adjective group or single adjective
adverbal = adverb group or single adverb
These cover terms are to be understood as form terms, notfunction terms.
Many grammarians go one step further and conflate nominal and pronom-
inal constituents in one broad category of 'nominale', the reason being that
the two types of constituent are very closely related, both functionally and
referentially. While not disputing the soundness of this approach in principle,
it is practical nonetheless to retain the term 'pronominal' and deal with nom-
inals and pronominale separately (in chapters 10 and 11, respectively).
3.3.2. The preposition group

As we have seen, the preposition group is special: it is not an expansion of a
single word and the relationship between H and DEP is different from that in
other groups. The preposition group is always binary, i.e. it consists of two
parts only, a head and one dependent. The dependent is traditionally referred
to as the 'prepositional complement'. The complement may be realized by a
nominal, a pronominal, a compound unit, or a nonfinite -ing clause:
(1) (The clever girls objected) A:ß[to the proposal]
H:prep[to] DEP:g[me proposal]
(2) (No, I haven't heard) A:g[from her]
H:prep[fr0m] DEP:Pro[her]
(3) (Richard flew) A:6[to Rome and Athens]
ttpreptto] DEP:cu[Rome and Athens]
A:
(4) 6[By leaving so early] (he offended my wife)
H:prep[By] DEP:cl[leaving so early]
The complement may also be realized by an interrogative clause (as in (5)) or

by a so-called independent relative clause (as in (6)), cf. sections 8.3 and
(5) I was wondering about which doll to give her.

(6) You can give this book to whoever wants to read it.
Colloquially, finite clauses are used after How about:
(7) How about I take Jenny to her music lesson?
Apart from examples like (5) to (7), finite clauses and nonfinite infinitive or
past participle clauses cannot serve as prepositional complements.
Cases like the following present a slight complication:
(8a) Before midnight, all the guests had disappeared.
(8b) Before he left, he thanked me profusely.
(9a) After breakfast, everyone joined the game.
(9b) After she arrived, everyone seemed far more relaxed.
In the a-examples we seem to have straightforward examples of prepositional
groups. In the b-examples, the status of Before and After is more debatable.
They seem to have the same meaning as in the a-examples. But since
prepositions are in general incapable of taking finite clauses as their
complements (e.g. *By he left so early he offended my wife}, it seems more
reasonable to analyse before and after in examples like (8b) and (9b) as
subordinating conjunctions (like e.g. when in When he left, he thanked me
profusely / * When midnight, all the guests had disappeared). In other words,
we consider items like before and after to be in a grey zone between
prepositions and conjunctions. For practical reasons we shall treat them as
prepositions when they are followed by (pro)nominals or present participle
clauses and as conjunctions when they are followed by finite clauses.
Another, very similar problem is provided by the following examples:
(10) She told him nothing except that Robert would soon join them.
(11) The case was very complicated in that so many firms were involved.
While some grammarians treat except and in as exceptions to the rule that
prepositions cannot take finite thai-clauses (cf. e.g. Schibsbye 1970: 189),
others regard the sequences except that and in that as complex subordinators
on a par with as if, in case and in order to (cf. Quirk et al. 1985: 998).
Preposition groups typically have either adverbial function at clause level
(as in example (12)) or dependent function in (pro)nominal groups (as in
example (13)):
(12) On Jack's advice she will fly to Rome.
(13) Someone from our office tapes the constant stream of interviews she gives.
The relationship between H and DEP in a preposition group is such that both
constituents are in fact more independent than in other groups. Note in this
connection the following points:
(i) The prepositional complement is occasionally capable of serving as the
subject of a passive construction, leaving the prepositional head 'stranded'
(cf. also sections 4.3.2 and 7.4.3):
(14a) Someone has slept in that bed.
(14b) That bed has been slept in.
(15 a) We must fight for freedom.
(15b) Freedom must be fought for.
In the b-examples the prepositional complements of the a-examples have
undergone a functional transformation and become syntactic subjects.
(ii) The prepositional complement is often separated from the head
preposition, thus realizing a discontinuous group:
(16a) You can draw water/ram this well.
(16b) This well you can draw water from.
In such examples, the prepositional complement does not undergo a
functional transformation but keeps its dependent status (for further
discussion, see section 5.6.4).
(iii) The prepositional complement may by itself undergo relativization, i.e.

be represented by a relative pronoun:
(17a) The professor referred to this passage.
(17b) This is the passage to which the professor referred.
(17c) This is the passage which/that the professor referred to.
In connection with relativization of the complement the preposition group
may be realized either as a continuous or discontinuous group, as in (17b)
and (17c) respectively.
(iv) The prepositional head may grow more closely related to a preceding
verb, forming a so-called 'prepositional verb' (cf. section 4.3.2):
(18) Alfred's wife always stood by Jack.
(i.e. 'Alfred's wife always supported Jack')
(19) Miranda waited on the Wilson family,
(i.e. 'Miranda served the Wilson family')
If the preposition is here analysed as part of the predicator, the 'prepositional
complement' (Jack and the Wilson family, respectively) assumes direct object
function.
It is important to note that many grammarians hesitate to speak of a head-
dependent relationship at all in the preposition group, arguing that the
relationship between the preposition and the prepositional complement is
more like the interdependent relationship between sentence constituents
(such as S and P or P and O), in that the two parts are normally both
obligatory and, though functionally different, seem equally important in
status. Preposition groups are thus in a sense more clause-like than group-
like and can be viewed as a kind of'contracted clause' (cf. Halliday 1994:
212ff). Note in this connection constructions like concerning your
application (cf. about your application), where the present participle serves
as a preposition-like item in a construction which basically seems to involve
a P O structure. In the sentence analysis system proposed here, however, con-
stituents consisting of a preposition and its complement are classified as
groups with the preposition as head because, like the heads of other groups,
the preposition is the characterizing element which distinguishes it from
other groupings of words. At the same time, however, we recognize the fact
that in preposition groups the relationship between Η and DEP is of a special
kind (for further discussion, see section 6.3.3 below).
Some grammarians hold that preposition groups are really noun groups
with the noun or nominal construction as head rather than the preposition, the
argument being that the preposition is a function word just like e.g. the
definite article in traditional noun groups. A possible counter-argument is

that a noun group with a definite determiner as dependent may still be a noun
group if we leave the definite article out, and it may still assume the same
kind of function in the immediate linguistic context (e.g. The clever girls
objected to the proposal vs. Clever girls objected to the proposal, where both
the initial constituents are clearly noun groups and both serve as the subject
of the sentence). But a preposition group ceases to be a preposition group if
we leave out the preposition, and what is left of the construction is usually
incapable of functioning like the original construction in the linguistic
context (e.g. / went to Paris —> */ went Paris). Moreover, in a preposition
group, it is the preposition, not the prepositional complement, which is the
formally stable element in the construction, the prepositional complement
being potentially realized by a number of different forms (cf. examples (1) to
(7) above).
3.3.3. The compound unit

Like groups, compound units may realize sentence functions:
S:cu
(1) [Rolf and Werner] (were devious devils)
(2) (Bernard was) C:cu[angry but calm]
(3) (He gave) Oi:cu[Tessa or Fiona] Od:cu[guns and bullets] A:cu[in London and
in Berlin]
A compound unit typically consists of two or more conjoints (abbreviated as
CJT) bound together by a coordinator (abbreviated as CO). Thus the internal
functional structure of the examples above looks like this:
CJT
(4) [Rolf] co[and] CJT[Werner]
CJT
[angry] co[but] CJT[calm]
etc.
Conjoints may be realized by any of the four types of form (word, group,
compound unit, clause); coordinators are realized by conjunctions:
(5) Wendy and Kim (sat round the kitchen table).
Kim
(6) (They saw) your daughter and my son (at the party).
cu
CJT CO CJT
g conj
your daughter and my son
(7) (The two winning teams were) Walter and Sally and Peter and Helen.
cu
CJT CO CJT
conj
Walter and Sally and Peter and Helen
(8) (I thought) that Jack had already left and that Helen would be coming to see
me.
that Jack ...left and that Helen ... me
The examples listed above are all examples of linked coordination, where the
conjoints are explicitly connected by a coordinator. Sometimes we get
unlinked coordination, where there is no overt coordinator (as in example
(9a)), but where a coordinator could be inserted (as in example (9b)):
(9a) Who blew the landing party, the coordinates, the beach, the timel
(9b) Who blew the landing party, the coordinates, the beach and the timel
In linked coordination of more than two conjoints, usually only the last two
conjoints are separated by a coordinator (as in example (9b)). However, all
the conjoints may be separated, as in examples like the following where there
is emphasis on each of the conjoints:
(10) It was only too easy to mistake our tortuous structures of codenames and
symbols and cutouts for life on the ground.
Coordination is examined in more detail in section 6.2.
3.3.4. The clause

Functionally, many clauses differ from sentences in being bound, i.e. in
being incapable of occurring on their own. Formally, on the other hand, they
are like sentences, for their internal structure is basically identical with that
of sentences. In clauses we find all the functions identified at sentence level:
subjects, predicators, objects, complements and adverbials. In addition, there
is sometimes a formal marker of subordination, a subordinator (abbreviated
as SUB) realized by a conjunction (conj), as in the following examples:
(1) (They figured) that Jack was a double agent.
that Jack was a double agent
(2) Whether he talked her into it (I simply do not know).
Whether talked her into it
Like sentence functions, clause functions may be realized by single words,

groups, compound units or clauses:
(3) (Richard suspected) that Jack would tell Ursula and me if he got the chance.
O.
cl
SUB S. P. O_ A
conj n g cu cl
that Jack would tell Ursula and me if he ... chance
It is customary to distinguish between main clauses and subordinate clauses

(henceforth referred to as subclauses). The main clause corresponds to the
whole sentence, including subclauses. Subclauses function within main
clauses, either by assuming a clause function (as in the examples above) or
by assuming some lower-level function (e.g. as a dependent in a group or a
conjoint in a compound unit). A further distinction which is sometimes
useful is that between main clause and matrix clause. The term matrix clause
is used about a superordinate clause minus its subclause (especially if the
subclause functions as object or adverbial in the main clause). In the
examples above, the matrix clauses are They figured, I simply do not know
and Richard suspected, respectively.
An important difference between main clauses and subclauses is that while
main clauses (and sentences) are finite, containing a finite predicator, sub-
clauses are often nonfinite, containing an infinitive or participle as predicator
(cf. section 3.2.1). Here are some examples of nonfinite subclauses:
(4) To see her is to love her.
(5) Having finished my homework, I went out.
A clause consists of at least two clause functions. As we have seen, there is
almost always a predicator in a clause. But the presence of a predicator is not
actually criterial for the definition of clauses. Occasionally we come across
'predicator-less' clauses, i.e. clauses where the predicator (and sometimes
also other clause functions) is implied, and where the remaining constituents
can only be analysed appropriately in terms of clause functions (cf. section
8.8):
(6) (I am sure he will help you,) A:clj·jf necessary]
SUBxonjjifJ C:adJ[necessary]
3.3.5. Embedding
In many of the examples cited in the preceding sections we have seen that the
complex forms (groups, compound units, clauses) realizing sentence func-
tions contain internal functions which themselves are sometimes realized by
complex forms, which then in turn require further analysis. Thus in example
(3) in section 3.3.4 (Richard suspected that Jack would tell Ursula and me if
he got the chance), we have functions within the Orel realized by all three
complex form types: group-within-clause (would tell), compound unit-
within-clause (Ursula and me), and clause-within-clause (if he got the
chance). There are in principle no restrictions on the constellation of form
types within form types or on the number of constituent levels in a sentence:
Constellations of complex forms within complex forms are often referred
to as embedding. The phenomenon that language in principle permits infinite
embedding is referred to as recursiveness because the rules of embedding
may recur at any constituent level. Since embedded constituents are of the
same form types as non-embedded constituents we can simply describe
embedded constituents the way we do non-embedded constituents, as in the
analysis of example (1) below, and we thus already have the full system for
describing sentences irrespective of length and complexity.
(1) Jack knew a young doctor who graduated from a university, where ...
Sent
_S P_ Od
n v g
Jack knew PEP PEP PEP

art adj cl
a young doctor S _P _A
pro v g
who graduated H PEP

prep
from PEP JI PEP

art n cl
a university where
4. Advanced sentence analysis
4.1. Stacking
4.1.1. Form stacks
We shall now examine a number of more complex syntactic issues and show
possible ways of dealing with them. The problem that we turn to first
concerns the nature of syntactic relations and their representation. So far we
have represented constituents hierarchically in terms of a consist-of rela-
tionship without concern for the more precise nature of the relationship
between individual constituents on the same level of analysis. For example,
in a subject noun group such as the following:
(1) ^-8[The beautiful little flower which she gave him] (was pink)
a simple consist-of analysis yields:
(2a)
PEP PEP PEP _H PEP

art adj adj n
The beautiful little flower which she gave him
In this representation, the dependents are not only placed on the same level in
relation to each other but they are placed on the same level as the head of the
group. Thus the analysis does not reveal that the dependent article
determines the head while the dependent adjectives modify the head, or that,
somehow, the head is a more important constituent than the dependents,
except through the fact that it is symbolized as H. Structurally, they are given
the same status. Notice also that the configuration is identical to that assigned
to sentence and clause constituents (subject, predicator, object, etc.), despite
the fact that very different relationships are involved here. Our intuition
about these different relationships is not reflected directly in the structuring
of constituents (but only to some extent in the terms used: e.g. 'dependent
article' vs. 'dependent adjective' vs. 'head noun'). For this reason it is often
tempting to offer a more detailed hierarchical structure, as in:
72 Advanced sentence analysis
(2b)
J3
g
PEP H
art χ
ι
The PEP
——-~ Η
beautiful -
adj
little H_ PEP
n d
flower which she gave him
The form 'x' here represents what we refer to as a form stack. A form stack is
a collection of constituents belonging together in one way or another without
obviously constituting one of the three basic complex form types already
introduced (clause, group or compound unit). In analysis (2b), the form stack
χ represents three progressively smaller units. These units resemble 'full
groups' (in having an internal DEP - H or H - DEP structure) but are, in this
particular example, only part of a group and therefore structurally limited
(e.g. by not accepting a determiner such as the definite article, which is
already present at a higher level in the group).
By contrast, the analysis in (2a) is intended to be fairly neutral: it simply
describes the subject group as a constituent which contains four dependents
and a head without stating the finer relationships involved. The analysis in
(2b) makes a stronger claim: not only are there four dependents and a head in
the subject group, they are related in terms of progressive subordination.
The motivation for sometimes choosing the more elaborate representation
in (2b) instead of the more neutral one in (2a) is the desire to show certain
important relationships or linguistic phenomena. We cannot say that the
analysis in (2b) is better or more correct than that in (2a); but being more
detailed, the analysis in (2b) may serve the purpose of showing something
about the subject group which is not captured in the analysis in (2a). What
(2b) may serve to reflect is the intuition that the dependents do not modify
the head independently. For example, it could be argued that beautiful does
not simply modify flower but little flower which she gave him, and that The
determines everything which follows it, not just flower.
Stacking 73
It is important to note that even the more elaborate structure in (2b) does not
reveal everything there is to say about The beautiful little flower which she
gave him. For example, there is a sense in which the definite article The is
more closely related with the dependent clause which she gave him than with
the other constituents. The definite article and the dependent clause could be
seen to jointly determine flower, the clause is intended to provide informa-
tion which is familiar to the listener, thus helping him or her to identify the
referent of flower. In other words, the clause has a specifying function and
thus in fact warrants the use of the definite article. There is no simple or
natural way of showing this kind of relationship in a tree diagram. While
obviously useful as a framework for discussion and as a first approximation
to the structure of a sentence, constituent structure (as displayed in e.g. tree
diagrams) has its limitations.
The use of form stacks in examples like (2b) is an attempt to attune our
sentence analysis to some of our intuitions about constituent structure. It
does not, however, enable us to show all the grammatical facts.
4.1.2. Function stacks: predicates and predications

As we have seen, it is sometimes useful to operate with form stacks to
represent a collection of constituents somehow belonging together without
constituting one of the complex form types already established in our system:
clause, group or compound unit. But it is also sometimes useful to operate
withyu/ic/io« stacks. A function stack, for which we use the abbreviation 'X',
is a collection of constituents somehow belonging together without
constituting one of the functions already established in our analysis (subject,
predicator, object, etc.; head, dependent, conjoint, coordinator). One such
function stack which many grammarians operate with is the so-called
predicate, which contains the predicator plus subsequent sentence or clause
functions (such as objects, complements and adverbials, but not subjects), as
in the following example:
(1) Barbara seized a plate.
Sent
Barbara
seized a plate
The predicate stack here includes the predicator and the object. By placing
these two functions in a function stack, we can indicate that together they
express some comment about the subject of the sentence (see section 3.2.2
on the topic/comment distinction). Note that the form stack which realizes
the predicate stack is here similar to a 'full clause* (in the sense that it can be
analysed in terms of clause functions like P and O) but cannot be said to be a
clause in its own right: it is only part of a clause and therefore structurally
limited (e.g. by not allowing a subject, which is already present at a higher
level in the clause as a whole). In this respect, the form stack is here like the
form stacks in example (2b) in section 4.1.1, which were not groups in their
own right.
While stacking is an optional refinement in cases like (1), it is difficult to
avoid in examples of coordinated predicates:
(2) Barbara seized a plate and gave it to Jack.
Sent
£5
n cu
CJT CO CJT
Barbara conj
and
gave it to Jack
The predicate is only one of a number of useful stacks. Consider the

following example (discussed in Bache et al 1993: 93f):
(3) He might drop into a bar.
This example could be analysed in a neutral way as consisting of a subject
pronoun (He), a predicator group (might drop) and an adverbial group (into a
bar). However, if one wants to capture the intuition that somehow the modal
verb might relates not only to the full verb drop but in fact to the rest of the
predicate (cf. 'What he might do is drop into a bar'), then this can be shown
in a tree diagram where we use a function stack for the full predicate and a
form stack for the predicate minus might. The modal verb retains its DEP
status but the Η status of drop is transferred to the form stack as a whole:
Stacking 75
Sent
He
might
drop into a bar
Again, while stacking is an optional refinement in simple cases like (3), it is

difficult to avoid in examples involving coordination:
(4) He might drop into a bar and down some liquor.
What we see coordinated here are not predicates but rather predicates minus
the operator. The stack consisting of a predicate minus its operator is
sometimes referred to as the predication. In He might drop into a bar and
down some liquor, two predications are coordinated: drop into a bar and
down some liquor. The stack convention allows us to illustrate the relation-
ships involved:
Sent
s _X
pro g
He DEP
cu
might CJT CO CJT

conj
A
— and
g
drop into a bar down some liquor
Stacking is also useful in examples like the following:

(5) Jack gave Barbara a kiss and Ann some good advice.
Sent
Jack
Barbara a kiss Ann some good advice
In this example, in which - for simplicity's sake - we do not operate with a

predicate stack, the function stack X represents a unit for which we have no
other name (not even a traditional one like 'predicate1 or 'predication'). The
form stack χ contains in each conjoint an indirect and a direct object. For
further discussion of complex coordination, see section 6.2.3.
The stack convention also allows us to show the difference in clausal
integration between adjuncts, on the one hand, and disjuncts and conjuncts
on the other (cf. section 3.2.9):
(6) Bill saw Jane clearly, (adjunct)
Sent
Bill saw Jane clearly
(7) Clearly Bill saw Jane, (disjunct)

Sent
X_
adv χ
I
Clearly J5 Ο
η η
Bill saw Jane

Stacking 77
(8) Bill saw Jane, anyway, (conjunct)

Sent
anyway
Bill saw Jane
As these tree diagrams show, adjuncts are closely integrated constituents at

clause level, while disjuncts and conjuncts are more peripheral: they relate to
the rest of the sentence as a unit. The relative closeness of the constituents in
this unit is captured by analysing them as belonging together in a stack.
4.1.3. A final comment on stacking

Tree diagram configurations of the kind employed in our sentence analysis
allow us to show a consist-of relationship between constituents but do not
reveal the nature of the relationship between constituents at the same level.
In part this can be remedied by using well-defined terms for the forms and
functions of constituents. Another useful technique is to use form and
function stacks. These allow us to place constituents at different levels of
analysis according to how closely related they seem to be. Stacks are
especially useful in the analysis of progressive subordination and of co-
ordination of complex units which do not clearly belong to the basic form
and function types with which we operate. A stack is used whenever we have
such a functionally or formally 'unusual' constituent.
As we have seen, there are in some examples several alternative analyses.
This is hardly surprising: in sentence analysis we attempt to map our
intuitions about language structure in a principled manner, and there is
inevitably an element of interpretation which makes the analysis subject to
an evaluation in terms of what is possible or not possible (or what is
appropriate or inappropriate), rather than what is right or wrong. There is,
moreover, an element of analytic purpose: to some extent our choice of
structural representation in any given case depends on what we want to show
about language, and on our hypotheses about language. Finally, even with
the sophistication provided by the stacking technique, constituent structure
does not reveal all the relevant grammatical facts.
4.2. Missing constituents, ellipsis and pro-forms

4.2.1. The zero convention
In sentence analysis, we occasionally get a sense of missing constituents. An
example of this is: Jack became president and Ann vice-president. This
sentence is a 'short version' of Jack became president and Ann became vice-
president. To avoid repetition of became we may well use the short version.
In the following, we shall examine this and other cases of missing con-
stituents and propose a way of handling them.
The reason why we sometimes get a sense of missing constituents is that
we have certain expectations as to the well-formedness of constructions.
These expectations are based on general, typical ways of matching meaning
with expression. Thus, normally, to convey the meaning 'Ann became vice-
president' we have to say just that: Ann became vice-president. When in a
specific linguistic context (such as Jack became president and ...) the
meaning of this construction is conveyed by the shorter expression Ann vice-
president, we sense that the predicator of the 'normal1 expression, became, is
somehow missing from the second half of the sentence. When a constituent
'goes missing' in this way it is customary to talk of ellipsis. Ellipsis is a
common device for abbreviating sentences and serves to avoid redundancy.
One way to deal with ellipsis in our sentence analysis is to incorporate the
appropriate function label in the 'place' of the missing constituent and to have
the function 'realized' by a 'zero form' (represented by '0') to indicate the lack
of formal expression. This approach allows us to treat the sequence Ann vice-
president as a clause consisting of a subject noun (Ann), a predicator that is
missing ('0') and a complement noun (vice-president):
(1) Jack became president and Ann vice-president.
Sent
CJT CO CJT
conj
Jack became president Ann (became) vice-president
Here are some other examples of missing forms:

Missing constituents, ellipsis and pro-forms 79
(2) We parked Ildiko's car behind Sander's.

Sent
We parked PEP II
n n prep g
Ildiko's car behind PEP H

n
Sander's (car)
(3) She told me she wanted to return.

Sent
She told me SUB S P^ O^

0 pro v g
(that) she wanted to return
(4) Francis never received the letter Ildiko sent from Rome.
Sent
Francis never received
«he letter O. J5 ,Ρ
0 η ν
(that/which) Ildiko sent from Rome

In example (2), the speaker again suppresses repetition of a constituent (car).

Note that if in this example we do not operate with a missing form, the
analysis cannot show the parallel functions of Ildiko's andSandor's. In
example (3), there is no suppression of repetition: here the explicit marker of
the subclause, the subordinator conjunction that, has been left out. In the
linguistic context of the example (She told me ...), this does not, however,
change the status of the clause, which is still an object subclause. Finally, in
example (4) the relative pronoun that or -which has been left out. Notice here
that the zero convention allows us to supply an object to the predicator verb
sent, which is quite obviously used in a transitive sense and thus can be
expected to have an object.
While the zero convention can be regarded as an optional refinement in
examples (1) to (4), there are cases where it can be used not only to reflect
our intuition that something is missing but to clarify the analysis of the overt
constituents in a significant way. A good example of this is found in
'discontinuous' constructions like the following:
(5) The scheme they spoke so enthusiastically about has been dropped.
Sent
PEP they spoke so enthusiastically

0 prep
(which/that) about
The advantage of employing the missing constituent analysis in cases like (5)
is that the preposition about is not left stranded: without the prepositional
complement we would be forced to analyse it as an A:prep or A:adv with no
obvious connection with the head noun scheme. In the analysis above, this
connection is captured by having 0 as a prepositional complement in place
of an overt relative pronoun representing scheme.
The picture that emerges from the examples provided above is that the zero
convention is a useful tool which allows us to reflect the intuition that one or
more constituents are missing. In each case the missing constituent can be
retrieved without this resulting in a change of the meaning or of the overall
syntactic organization of the sentence. And in some cases the use of a zero
constituent in the place where the missing form is felt to belong helps to
clarify the relations between the constituents of the sentence.
4.2.2. Types of ellipsis

In the examples discussed in section 4.2.1, missing forms can be explained in
terms of 'economy of expression* within the sentence: we can retrieve the
missing forms by looking at other constituents in the same sentence. Thus, in
Jack became president and Ann 0 vice-president the missing predicator in
the second conjoint is retrievable from became in the first conjoint. And in
She liked the story 01 told them about Jack, the missing form that or which
has to be an appropriate relative pronoun representing story. In some cases,
the retrievability of a constituent is dependent on our grammatical know-
ledge of the relationships between the constituents in the sentence rather than
on the presence of a particular constituent. An example of this is She told me
0 she wanted to return, where the SUB:conj that is retrievable on the basis
of the subordinate status of the clause she wanted to return in relation to the
clause She told me. Similarly, in examples like the following:
(1) Serves you right.
(2) Anything the matter?
the hearer is capable of inferring the missing words too, not because of the
linguistic or situational context but because he knows that the utterances are
in free variation with It serves you right and Is anything the matter? All the
examples of missing forms discussed so far display ellipsis with intra-
sentential retrievability, or 'intrasentential ellipsis' for short.
In addition to intrasentential ellipsis we often come across missing forms
that can only be retrieved from outside the sentence, either from the
linguistic context, i.e. from other sentences, or from the situational context in
which the communication takes place. Consider the following examples:
(3) Speaker A: Bill came at eight.
Speaker B: Susan already at seven.
(4) Speaker A: I thought you were in a biology department.
Speaker B: I was.
We can argue here that the predicator came in (3) and the adverbial in a
biology department in (4) have been omitted. In both (3) and (4) the missing
constituent is easily recoverable from the preceding linguistic context. What
we find here is thus extrasentential ellipsis, i.e. ellipsis with extrasentential
retrievability. Such ellipsis is not restricted to examples of communicative

turn-taking:
(5) Quite frankly, I left this morning. 0 Couldn't stand the noise.
The missing subject can be retrieved as / from the preceding sentence.
In some cases the part omitted from a simple sentence is not recoverable
from the linguistic context but from the situational context (see Greenbaum
and Quirk 1990: 256f):
(6) Told you so.
(7) See you later.
In the absence of a clarifying linguistic context we cannot be sure whether it
is / and /'// or We and We'll which are missing. In a specific situational
context, however, the hearer will often be able to determine precisely what is
missing, so in examples of this type it seems warranted to operate with
situational ellipsis.
Ellipsis can be described as anaphoric, cataphoric or nontextual according
to how we retrieve the missing form(s). Retrievability which depends on the
presence of some constituent in the linguistic (intra- or extrasentential)
context rather than simply on our grammatical knowledge, is typically ana-
phoric: the position of the missing element is after the position of the
constituent that allows us to retrieve the missing element. Examples like Jack
became president and Ann 0 vice-president and Quite frankly, I left this
morning. 0 Couldn't stand the noise display anaphoric ellipsis. Cataphoric
ellipsis is rare. Arguably we have it in examples like Jack's 0 is a beautiful
voice and She handed me the red 0 and the yellow box. Nontextual ellipsis
can be illustrated by (6) and (7) above, as well as examples like e.g. She told
me 0 she -wanted to return.
Ellipsis can be further described in terms of the position in the clause of the
missing constituents). In examples like 0 Serves you right and 0 Couldn't
stand the noise, we have initial ellipsis. The omitted element(s) here would
typically have constituted the prehead of a tone group, i.e. would have been
pronounced without stress and pitch prominence. This lack of prosodic
prominence reflects their paucity of information value and sheds light on
their susceptibility to ellipsis. In constructions like Jack became president
and Ann 0 vice-president and Susan 0 already at seven, we have medial
ellipsis. Medial ellipsis is typically the result of the suppressed repetition.
Finally, it can be argued that we have terminal ellipsis in examples like I was
0 and We parked Ildiko's car behind Sandor's 0. In this position, too, ellipsis
is usually the result of suppressed repetition.
4.2.3. Pro-forms
Ellipsis is not the only way of achieving economy of expression. Consider
the following examples:
(1) {My little sister} saw herself in the mirror.
(2) They arrested {Jeremy Soames}, who was on his honeymoon.
(3) When {Jill's mother) asked for {a new film for the camera}, I gave her one.
The italicized constituents in these examples are pro-forms, more specifically
pronouns, representing other constituents in the sentences, viz. those appear-
ing in curly brackets. In (1) the reflexive pronoun Aerse/frepresents My little
sister, in (2) the relative pronoun who represents Jeremy Soames, and in (3)
the personal pronoun her represents Jill's mother and the indefinite pronoun
one represents a new film for the camera. In all four cases, the pronouns are
light constituents standing proxy for heavier constituents in the linguistic
context. Like ellipsis, the use of pro-forms secures economy of expression:
rather than repeating a heavy constituent we use a light pro-form.
Most pro-forms present no difficulty with respect to sentence analysis. In
the examples above, each pronoun assumes an easily identifiable function:
herself is O:pro, who is S:pro, her is Oi:pro and one is Od:pro.
Certain pro-forms may represent larger chunks of material in the linguistic
context:
(4) He's a friendly dog called Poulidor, but {he's now got so old that he's gone
stone deaf}. Both Oliver and I und this terribly sad.
(5) Deny it though he might, {he dumped his wife in Paris}.
In these examples, the pro-forms represent a whole clause, this anaphorically
and it cataphorically. Again the analysis is fairly straightforward: both pro-
forms are O:pro. There is a tradition for treating items like this and it as
pronouns even in cases where they represent larger constituents, such as
clauses.
Sometimes pro-forms represent less clearly identifiable constituents:
(6) He may decide to join us next week, but I don't think so.
(7) Speaker A: Will he join us next week?
Speaker B: I hope not.
In (6) the adverb so represents the preceding clause but with an appropriate
change of operator: ... but I don't think he will join us next week. In (7), the
adverb not represents the whole of the preceding clause but changes it into a
negative statement: ... / hope he will not join us next week. The proposed
analysis of both items is O:adv.
Consider next the following examples:
(8) I would like to go to Rome. So would my wife.

(9) He cannot withdraw at this point. Nor can she.
(10) My father didn't like Jenny, and neither did my mother.
Here we have complex pro-forms consisting of an adverb (so, neither, nor)
plus an operator (either a repeated operator or a form of D O). In each
example, the combination as a whole represents a predicate stack:
(8?) ... My wife would like to go to Rome too.
(9") ... She cannot withdraw at this point, either.
(10') ... and my mother didn 't like Jenny, either.
The analysis of such examples is not entirely straightforward. In each case,
the adverb stands proxy for a predication stack, i.e. a predicate minus the
operator (e.g. like to go to Rome too). We therefore propose to analyse the
combination of the operator and the adverb in the same way as we analysed
the combination of an operator and a full predication in section 4.1.2, i.e. in
terms of a DEP - Η relationship within a predicate stack:
Sent
would
Finally it may be noted that it is not always easy to decide whether an

example shows ellipsis or the use of preforms:
(11) For years he wanted to win the race and he finally did.
(12) Speaker A: Will he propose marriage to her!
Speaker B: Well, he may.
In these examples, the operator (did, may) either serves as a pro-form
representing a whole predicate (won the race and may propose marriage to
her) or is left stranded after ellipsis of the predication (win the race, propose
marriage to her). Similarly, in an example like We parked Ildiko's car behind
Sandor's (cf. section 4.2.1), an alternative to saying that there is ellipsis of
car after Sandor's is to regard Sandor's as a pro-form for Sandor's car.
Complex predicators 85
4.3. Complex predicators

4.3.1. Phrasal verbs
So far, all the predicators we have looked at have consisted of verbs only. In
this and the following sections we shall deal with predicators that are
somewhat more complex in that they consist not only of verbs but also of
items from other word classes. These items have become so closely related
with the full verb that they are felt to make up an integrated unit of
expression with it. We turn first to predicators in which the full verb is fused
with a following adverb, the so-called 'phrasal verb':
(1) Thor turned down the generous offer.
(2) Julia called up Simon.
(3) Cassandra gave in eventually.
We analyse the predicators in examples like (1) to (3) as consisting of
phrasal verbs because in each case the verb and the following adverb are felt
to constitute a formal and semantic unit. To appreciate the close relationship
between the adverb and the full verb in a phrasal construction, we can
compare example (1) to the following:
(4) Thor turned down the street.
(5) Thor went down to Melbourne the other day.
In example (4), turned has the fairly concrete meaning of'changing direction'
(unlike turned in example (1), which seems to derive a new, more abstract
meaning in the phrasal construction) and down is clearly a preposition
relating to the street, indicating 'direction' and 'path'. In example (5), down
can be viewed as an adverb functioning separately as an adverbial specifying
the direction of the event expressed by went. In both instances down is felt to
be much more independent of the full verb than in example (1).
The combination of a verb and an adverb in a phrasal construction can
usually be replaced by a simple verb with a similar, or identical meaning:
(1') Thor rejected the generous offer.
(2*) Julia phoned Simon.
(3') Cassandra surrendered eventually.
In the clearest cases of phrasal predicators, the verb and the adverb are thus
fused semantically: the meaning is not simply a composite of the meaning of
the verb and the meaning of the adverb but rather derived in a fairly
unpredictable way. Neither verb nor adverb seems to retain the meaning they
usually have when used more independently. As we see in examples (1) to
(3), up, down and in have lost their concrete spatial or directional meaning;
and the relationship between call and phone, turn and reject, and give and
surrender is very tenuous.
There are several characteristic features of adverbs in phrasal verbs: a) they
are typically drawn from a fairly small set of mono- or disyllabic adverbs
(ABOUT, ACROSS, DOWN, IN, OFF, ON, OUT, OVER, UP, etc.); b) outside
the phrasal construction they have general locative (spatial or directional)
meaning; c) they are capable of serving also as prepositions (e.g. up the
chimney, down the street, in a bad mood); and d) they receive primary stress.
Because of their subordinate, dependent status in phrasal constructions they
are often referred to as 'particles'.
One syntactic characteristic of adverbs in phrasal predicators is that, unlike
many other adverbs, their position is relatively fixed. In intransitive phrasal
predicators, the adverb always immediately follows the full verb (as in
example (3) above). In transitive constructions with a direct object which is
not realized by a pronoun, the adverb may occur before or after the object
with no difference of meaning:
(la) Thor turned down the generous offer.
(Ib) Thor turned the generous offer down.
In pre-object position the full verb and the adverb are generally inseparable
in the sense that no other constituent may intervene:
(1 c) *Thor turned again down the generous offer.
If the object is an unstressed pronoun, the adverb follows the object:
(l d) Thor turned it down.
(1 e) *Thor turned down it.
In intransitive constructions and in transitive constructions with the adverb in
post-object position, the adverb bears the nuclear tone if it is the last word of
the sentence.
Transitive phrasal constructions allow of pronominal question form with
who or what like other S P O constructions (see section 3.2.5 on the identi-
fication of objects):
(1 f) What did Thor turn down!
It is not always possible to distinguish rigidly between phrasal predicators (as
in e.g. example (1)) and non-phrasal predicators (as in e.g. examples (4) and
(5)). Consider the following examples:
(6) The children put on the gumboots.
(7) Francis took away the little silver box.
(8) Cassandra hielt down.
In these examples, the full verbs put, took and knelt retain much more of their
independent meaning than the full verbs in the other phrasal constructions
reviewed in this section; in addition, the adverbs on, away and down have a
clearly locative meaning. Yet the transitive examples (6) to (8) conform very
nicely to the rule of adverb mobility applying to phrasal predicators, e.g.:
(6a) The children put the gumboots on.
(6b) The children put them on.
(6c) *The children put on them.
In the intransitive construction in example (8) (Cassandra knelt down),
although the verb and the adverb retain their independent meaning as in
clearly non-phrasal constructions, there is a close relationship between them:
the locative meaning of down is somehow integrated in the meaning of knelt.
We cannot, for example, say * Julia knelt up/away/across, etc. In this respect
knelt down is different from went down in example (5) (Thor went down to
Melbourne the other day): it is quite possible to say Thor went up/away/a-
cross etc.
There is thus a gradation of constructions from very clear cases of phrasal
fusion between verb and adverb to very clear cases of verb and adverb
functioning independently. In between these two extreme end-points there
are constructions which conform to some but not all of the criteria.
It is not easy to offer a fully satisfactory syntactic analysis of phrasal verbs,
simply because the phenomenon involved is primarily to do with lexical-
ization. But since in English this kind of lexicalization does not lead to a
formal fusion at word level, we are forced to try to cater for it syntactically.
The simplest, but also most unsatisfactory, way of analysing sentences with
phrasal predicators is to disregard the phrasal nature of the verb and adverb:
Sent
Julia called up Simon
If we want to reflect the phrasal combination of the verb and the adverb in
our analysis, this can be done by including the adverb as a dependent in the
predicator:
Sent
Julia Simon
called
If one wants to show the close relationship between verb and adverb while at
the same time keeping the predicator 'pure' (i.e. strictly for verbs only), e.g.
for identificatory purposes (see section 3.2.1), one alternative is to use the
stack convention, delaying the assignment of sentence functions:
Sent
Simon
called up
To each of the possible analyses of phrasal predicators there is a dis-

continuous counterpart for cases with the adverb in post-object position. We
leave it to the reader to work out what they look like! As always, the choice
of analysis among competing options very much depends on exactly what
you want to show and how.
4.3.2. Prepositional verbs

Predicators can sometimes be interpreted as containing a full verb followed
by a preposition with which it is felt to form a formal and/or semantic unit.
Such combinations of a verb and a preposition are traditionally referred to as
'prepositional verbs'. In the following examples it is possible to analyse the
predicator in such terms:
(1) Miranda waited on the Wilson family.
(2) Alfred's wife always stood by Jack.
(3) Stephen took after his father.
These can be compared with clauses where an ordinary predicator is clearly
followed by an adverbial realized by a prepositional phrase:
(4) Miranda waited on the comer.

(5) Alfred's wife always stood by the fireplace.
(6) This is what Stephen took after his divorce.
Notice that in these examples the full verb and the preposition have com-
patible but fairly independent meanings. By contrast, the predicator and the
preposition are semantically fused in examples (1) to (3) and could be
replaced by single verbs with little difference of meaning:
(Γ) Miranda served the Wilson family.
(2') Alfred's wife always supported Jack.
(3') Stephen resembled his father.
Unlike phrasal predicators, prepositional predicators are always transitive,
taking a direct object. This object is sometimes referred to as a 'prepositional
object' to indicate that, despite the analysis of the preposition as belonging to
the predicator, there is still a close relationship between it and the following
constituent. Like Ordinary' S P O constructions, sentences with prepositional
predicators allow pronominal question forms with who or what (see section
3.2.5 on the identification of direct objects):
(la) Who(m) did Miranda wait on?
(2a) Who(m) did Alfred's wife always stand by?
(3a) Who(m) did Stephen take after?
This obviously lends credence to the analysis of such combinations as
prepositional predicators in S P O constructions: unlike the analysis of
phrasal verbs, the analysis of prepositional verbs affects the function of the
following constituent. If we treat the preposition as a part of the predicator
(the 'prepositional predicator' analysis), the following constituent is an
object. But if we treat the preposition as outside the predicator (the 'simple
predicator' analysis), the following constituent is a dependent in an adverbial
preposition group. The fact that pronominal question form with WHO is
possible in the examples above indicates that we have an S P O structure
rather than an S P A structure.
Unlike adverbs in phrasal constructions, prepositions in prepositional
predicators are normally unstressed and constitute the tail of the tone group if
in clause-final position (e.g. Who did Miranda *wait on, where nuclear stress
is symbolized by a preceding accent mark). And while adverbs in transitive
phrasal constructions may occur in post-object position (if the object is not
pronominal), prepositions are fixed in pre-object position:
(lb) *Miranda waited the Wilson family on.
Another difference between phrasal and prepositional predicators is that it is

possible to separate the preposition from the verb even when the two are
closely fused semantically, either by inserting an adverbial between the two
or by forming a wA-question fronting both the preposition and the pronom-
inalized prepositional object:
(Ic) Miranda waited diligently on the Wilson family.
(Id) On whom did Miranda waif?
In some cases the verb and the preposition may be separated by another
object, which arguably yields ditransitive prepositional predicators:
(7a) The incident reminded me o/her warning.
(8a) Her stepparents deprived her of her childhood.
(9a) The countess supplied him with opium.
The verb and the preposition form a continuous group in passive construc-
tions with the non-prepositional object of the active construction as subject:
(7b) I was reminded o/her warning.
(8b) She was deprived o/her childhood.
(9b) He was supplied with opium.
Pronominal question forms with who or what are possible (cf. section 3.2.5
on the identification of objects):
(7c) What did the incident remind me ofi (her warning)
(8c) What did her stepparents deprive her of? (her childhood)
(9c) What did the countess supply him with! (opium)
As with phrasal predicators (see section 4.3.1 above), it is very difficult to
distinguish rigidly between prepositional predicators and predicators fol-
lowed by a prepositional group realizing a separate adverbial at clause level.
There is a clear gradation in the degree of semantic fusion between verb and
preposition: at one end of the scale both verb and preposition seem to lose
their independent meaning, forming a tightly knit unit with a derived, more
or less unpredictable, meaning (as in examples (1) to (3) above). At the other
end of the scale, we find examples where the preposition and its complement
quite obviously function together as an adverbial (as in examples (4) to (6)).
In between these two end-points, there are cases where there seems to be a
relatively close link between verb and preposition but little actual semantic
fusion, both constituents largely retaining their Original1 meaning. This is
true not only of the 'ditransitive' constructions offered above but also of
many 'monotransitive' constructions:
(10) Julia often looked at the picture.
(11) As always, I dealt with his request at once.

(12) My parents did not exactly approve of this decision.
Here looked, dealt and approve mean exactly what they say and the preposi-
tions at, with and o/are relational in meaning as we would expect them to be
when serving as heads in preposition groups. Semantically, there is thus
perhaps little motivation for treating the predicators in these examples as
comprising the preposition: they could simply be classified as S P A con-
structions (where the A is realized by a preposition group) rather than S P O
constructions (where the preposition is a part of the P). Yet, in each case, the
preposition is regularly associated with the verb (i.e. verb and preposition
usually 'go together') and together they often have nearly the same meaning
as non-prepositional predicators:
(10') Julia often saw the picture.
(11') As always, I treated his request at once.
(12') My parents did not exactly like this decision.
Possibly the best argument for analysing (10) to (12) in terms of pre-
positional predicators is that they can be passivized like other S P O
constructions:
(1 Oa) The picture was often looked at.
(1 la) As always, his request was dealt with at once.
(12a) This decision was not approved of.
Here the 'prepositional object' (strictly: the dependent in the adverbial
preposition group) in each of the examples in (10) to (12) has become the
subject of a corresponding passive sentence and thus behaves like an object
in an ordinary S P O sentence. Notice that unless we analyse the preposition
as a part of the predicator, it is left 'stranded' by passivization, its dependent
having assumed subjecthood, i.e. an independent function. A stranded
preposition is obviously difficult to analyse. But if we have the preposition
instead as a part of the predicator, it can be fully accounted for. The only
problem with this argument is that passivization, as well as the stranding of
prepositions, is possible in cases that we would not normally think of as even
remotely involving prepositional predicators:
(13a) Alice slept in the bed.
(13b) The bed was slept in (by Alice).
In such constructions, 'subjectivization' of a prepositional dependent seems to
be possible only when it is affected by the situation expressed by the
predicator (cf. The problem was gone into / *The tunnel was gone into).
Again we are forced to conclude that language is at times very elusive: we
cannot always capture its structure in a revealing way by applying absolute

criteria to what is in reality a gradation. It is important to recognize the
fuzziness of distinctions that become apparent in a rigid analysis.
Before concluding this section, we shall look at the various ways in which
we can use our sentence analysis system to show a close relationship
between verb and preposition. As with phrasal verbs, it is not easy to offer an
entirely satisfactory syntactic analysis of prepositional verbs, simply because
the phenomenon involved is primarily of a lexical nature. But again we are
forced to provide a syntactic representation: the lexical-like fusion between
verb and preposition does not lead to a formal fusion at word level.
There are two obvious competing analyses of clauses involving some sort
of relationship between a verb and a preposition: a) a basic one in which the
preposition is not syntactically recognized as part of the predicator but rather
is the head of an adverbial group containing the following constituent as a
dependent (the 'simple predicator1 analysis); b) one in which the preposition
is recognized syntactically as part of the predicator, viz. as a dependent
preposition, which leaves the following constituent as an object rather than
as a dependent in an adverbial group (the prepositional predicator analysis):
(14a)
Sent
Miranda waited H DEP

prep
on the Wilson family
(14b)
Sent
Miranda _H DEP the Wilson family

v prep
waited on
The analysis in (14a) is in keeping with the simplest possible analysis of

phrasal predicators mentioned in section 4.3.1. The only advantage of this
'basic1 analysis is that we keep the predicator 'pure' by only accepting verbs
as internal constituents (full verbs and auxiliary verbs) and by allowing no
dependents to the right of the head verb. In this way we gain a high degree of
internal consistency in predicators and make them more unambiguously
identifiable. The disadvantage is that we inevitably create 'external' com-
plexity in the sense that we make it more difficult to analyse adequately what
is left outside the predicator. For example, by making no syntactic difference
between Miranda waited on the Wilson family and Miranda waited on the
corner, we obscure the fact that, intuitively, the relations between the
constituents of these two sentences are very different, something that should
be reflected in our analysis.
If we choose the analysis in (14b), we may get more complex predicators
(by allowing non-verbal elements as internal constituents and by accepting
dependents to the right of the head verb). But the analysis enables us to
capture the difference between prepositional predicators and S P A construc-
tions. One other advantage of the analysis in (14b) is that we avoid stranding
the preposition in passive counterparts (such as The Wilson family was
waited on).
If we adopt the prepositional predicator analysis (as in (14b)) and want to
carry it through consistently, we are, on the other hand, forced to accept a
number of discontinuous constructions, e.g.:
(9c) What did the countess supply him with! (O P- S -P- O -P)
In this example, we might have assigned indirect object status to him and
direct object status to What. But in other ditransitive constructions with a
prepositional predicator, the assignment of direct and indirect object status is
more problematic (e.g. in She confined her remarks to the matter of income
tax). We therefore propose simply to leave the second object as an implicit
prepositional object.
Again, the choice of representational form is really a question of exactly
what one wants to show, and thus involves an element of subjectivity.
4.3.3. Phrasal-prepositional verbs

Sometimes there is fusion between a verb and both an adverb and a
preposition, as in the following examples:
(1) Cassandra looked down on the nurses.
(2) She came up with a solution in no time.
(3) He no longer put up with her whims.
The order of constituents in such constructions is always: verb + adverb +

preposition, reflecting the different degrees of separability in phrasal
predicators and prepositional predicators, respectively. To qualify as a
phrasal-prepositional verb, there should be some fusion between all three
elements with a new derived meaning; compare examples (1) to (3) with the
following, where the adverb and the preposition are more independent of the
verb, and all three elements actually retain their original lexical meaning:
(4) From the tower one could look down on the roof of Julia's house.
(5) After a couple of minutes the diver came up with an interesting shell.
(6) Did you read the notice he put up with yellow tape?
Notice also that phrasal-prepositional predicators can often be replaced by a
single verb with little or no difference of meaning:
( Γ) Cassandra despised the nurses.
(2') She found a solution in no time.
(31) He no longer tolerated her whims.
Arguably, the examples like the following involve 'ditransitive' phrasal-
prepositional constructions:
(7) Stephen always took her infidelity out on me.
(8) They put her behaviour down to lack of confidence.
Both 'monotransitive1 and 'ditransitive' phrasal-prepositional constructions
pass the wA-question test for objecthood like ordinary S P O constructions:
(1 a) Who did Cassandra look down on?
(8a) What did they put her behaviour down to"?
It is, of course, always possible to represent phrasal-prepositional con-
structions in terms of separate, non-integrated functions, in which case both
She came up with a solution and The diver came up with an interesting shell
are interpreted as consisting of a subject, a predicator and two adverbials (the
first realized by an adverb, the second by a prepositional group). If one wants
to recognize a syntactic difference between phrasal-prepositional predicators
and simple predicators, the simplest way of doing so is to have both the
adverb and the preposition as dependents within the predicators:
(9) His father had walked out on his mother.

Sent
His father PEP DEP DEP his mother

v v adv prep
I I
had walked
I
out
I
on
(10) They let her brother in on their plans.

Sent
S P- Ο -P Ο
pro g- g -g g
They H her brother

A
DEP DEP their plans
V adv prep
1
let in on
As with 'ditransitive prepositional predicators', the assignment of direct and

indirect object is not entirely unproblematic in 'ditransitive phrasal-
prepositional constructions'. We therefore leave the first object as an
ordinary, unspecified object as in a monotransitive construction, and the
second object as an implicit prepositional object.
4.3.4. Other complex predicators

A number of expressions, largely of an idiomatic nature, seem to involve
other combinations of words in predicators. Here are some of them:
v + adj e.g. CUT short, FORCE open, PUT straight, RUB dry,
BREAK even;
v + n (+ prep) e.g. TAKE place, BRING home, MAKE amends (for),
GIVE offence (to), PAY heed (to), PAY attention (to),
TAKE care (of), LOSE touch (with), CATCH sight of,
GIVE rise to;
v + v (+ prep) e.g. MAKE do with, GET rid of, HAVE done with, BE
going to;
others BE on the point of, BE about to.
In many of these cases it is possible simply to ignore the semantic fusion:

(la) S:pro[They] P:v[paid] O:n[attention] A:g[to his behaviour].
S: ro
(2a) P [She] P:v[was] A:8[on the point of leaving].
If one wants to indicate the close relationship between the units of these
more or less fixed expressions, the simplest possible way is to populate the
predicator with all sorts of non-verbal constituents closely related to the
lexical verb:
S: ro
(lb) P [They] P:8[paid attention to] O:8[his behaviour].
S: ro
(2b) P [She] P:ß[was on the point of leaving].
Semantically these analyses make good sense: PAY attention to corresponds
to HEED and BE on the point of leaving can be viewed as a special form of
the verb LEAVE, just like the progressive form in She was leaving. But it is
problematic to determine the internal structure of such fixed or idiomatic
expressions: how do we determine the H - DEP relationships in paid
attention to and was on the point of leaving? What exactly is the head of the
construction in expressions like MAKE do with, GET rid of ana BE going to?
The fixed nature of these expressions makes it hard to think of internal
relations and hence to analyse their constituent structure. This lack of
syntactic transparency, which is caused by the primarily lexical nature of the
phenomena involved, naturally opens up for a large number of competing
analyses, none of which is completely satisfactory.
Rather than pursuing the problem posed by fixed expressions, we shall
close this section with an observation on certain v + v (... + v) constructions
which pose a special problem of analysis. Consider the following examples:
(3) They wanted to leave the party early.
(4) Rosemary remembered posting the letter.
(5) Jack tried to stop smoking.
In these examples, we have strings of two or three verbs. Though it is tempt-
ing to treat them as complex predicators, there are good reasons why they
should be analysed in terms of single-verb predicators (wanted / remembered
/tried) followed by clausal objects (to leave the party early /posting the
letter / to stop smoking): all the verbs are full verbs according to the
definition provided in section 3.2.1 rather than combinations of auxiliaries
and full verbs; also, whatever follows the first verb can be identified as an
object in the wA-question test for objecthood:
(3a) What did they want? (To leave the party early)
(4a) What did Rosemary remember? (Posting the letter)
(5a) What did Jack try? (To stop smoking)
The top of the tree 97
Notice finally that the single-verb predicator analysis captures the similarity
between the examples in (3) to (5) and examples containing the same
predicator verbs but (pro)nominal objects:
(3b) They wanted it.
(4b) Rosemary remembered the whole thing.
(5b) Jack tried something else.
In these examples we would not hesitate to treat -wanted, remembered and
tried as predicators in their own right.
4.4. The top of the tree

4.4.1. Exit Sent
In this and the following sections we shall reconsider the top label Sent in
our analysis and replace it by a set of more revealing form and function
terms. As we saw in section 2.2, native speakers of a language may have
strong intuitions about what a sentence is and may be perfectly capable of
dividing a text into appropriate orthographical or intonational units.
Nevertheless, it is very difficult to define exactly what a sentence is. One of
the problems is that, even in writing, which is often considered the medium
susceptible of analysis in terms of sentences, punctuation is a very unreliable
criterion. Thus, as noted in section 2.2, we find units which we hesitate to
call sentences but which we treat like sentences punctuationwise:
(1) Spain.
(2) What?
(3) Good God!
In (1), Spain may, for example, be the answer to a question like Where do
you come from? In (2), we have possibly one of the most basic questions of
all in the form of just one word. And in (3), we have an exclamation, i.e. a
speaker's emotional reaction to some event, in the form of a simple two-word
group. The units in (1) to (3) are perhaps more characteristic of speech or
written representations of dialogue than actual written language. But even in
writing which is not meant to represent dialogue we often find units of text
which are not sentences in any obvious sense:
(4) An Introduction to English Sentence Analysis.
(5) Fragile.
(6) Bureaucratic battle over commas and spelling.
Here we have, respectively, a group serving as the title of a book, a one-word
label on e.g. goods to be transported, and a group serving as a news headline.
Obviously we want to be able to handle cases like (1) to (6) in our sentence
analysis system. To that end we need to revise the starting point of our
analysis, the Sent label. So far Sent, which is neutral with respect to the form-
function distinction (which we insist on at all other levels of constituency),
has been a convenient cover term for the kind of language unit we wanted to
analyse. But when confronted with a richer variety of speech and writing it
proves inadequate. It is inadequate from a formal point of view because it
signals the presence of a clausal structure, which is often simply not there.
As we have seen in (1) to (6), there are units of speech and writing which do
not fit into this form of structure but seem rather to be single words or
groups. Sent is also functionally inadequate because it does not allow us to
distinguish between the various functions that sentences and other
independent units seem to adopt in a larger context. Such functions, which
can be identified in both speech and writing, are not syntactic but commun-
icative functions, or "speech acts', such as e.g. questions or statements. So
what we want instead of Sent is a form-function distinction allowing a broad
range of both forms and functions.
4.4.2. Communicative functions

It is perfectly reasonable to begin by asking if communicative functions are
at all like the functions identified at lower levels of analysis (such as subject,
subordinator, head, etc.). At first blush they seem to be radically different:
while lower-level, grammatical functions seem to indicate relationships
between constituents, communicative functions like 'statement1 and 'question'
indicate a relationship between language and situation (in a broad sense).
For example, a question indicates that the speaker (or writer) wants to obtain
some information from the listener (or reader). However, there is some
similarity between lower-level functions and communicative functions in
that the language form to be assigned a function is related 'externally' to 'the
next level up'. For example, a subject is a function at clause level, a
dependent is a function at group level, and so forth. Such functions
characterize external relations of the constituents: a subject is a subject in
relation to a predicator, a dependent is a dependent in relation to a head, and
so forth. In a sense these lower-level functions license the occurrence of
constituents within higher-level clauses and groups. When we reach a
complete unit with a communicative function like a sentence (or a group or
individual word, as in examples (1) to (6) in section 4.4.1), there may or may
not be a next level up in terms of a higher-level formal unit, i.e. there may or
may not be a larger text within which the unit is a part. But in either case, it
is always possible to identify an external function of the unit in relation to
the context within which it occurs. This function justifies the occurrence of
the unit in the larger context (textual or non-textual), i.e. it is a commun-
icative function characterizing the speech act (or act of writing) producing
the unit. It is in this sense that a communicative function of a language unit is
like a lower-level syntactic function: it characterizes an external relation to
'the next level up' and thus licenses the occurrence of the unit. The difference
is that while the next level up for a lower-level function is always textual and
within the bounds of grammar, for a communicative function it is the (textual
or non-textual) context within which it occurs and strictly outside the bounds
of grammar.
The communicative function of a language unit is often referred to as the
illocutionary value of the unit. There is no obvious limit to the number of
such values that might be identified in speech and writing if one wants a very
fine-grained analysis. However, we shall here closely follow the tradition of
operating with four: statements (STA), questions (QUE), directives (DIR)
and exclamations (EXC). These four main illocutionary values, or speech
acts, will often, but not always, be marked differently in the internal syntactic
organization of the language unit or in its intonational properties:
STA (= statement): unit which gives information
(1) James left Brisbane yesterday.
(2) In London.
(3) John and Sarah.
(4) Yes.
QUE (= question): unit which seeks information
(5) Will you join me tonight?
(6) From whom?
(7) When and where?
(8) Why?
DIR (= directive): unit which instructs the receiver to perform some action
(9) Listen to me.
(10) After him!
(11) Smile and be happy!
(12) Down!
EXC (= exclamation): unit which indicates emotional reaction (surprise,
disapproval, pleasure, etc.)
(13) She can't mean that!
(14) Good Lord!
(15) Blood and sand!

(16) Wow!
Three other illocutionary values are obviously important but somewhat more
specialized and mostly restricted to formal expression:
PER (= performative): unit which actually performs the situation stated.
Performatives usually contain a 'performative verb' (e.g. PROMISE, SWEAR,
SENTENCE) in the present tense and a first person (singular or plural) agent.
(17) I (hereby) pronounce you man and wife.
(18) We (hereby) promise to support James.
OPT (= optative): (from a Latin word meaning 'choose, wish') unit which
expresses a wish or a benediction/malediction. Optatives often involve a
fixed expression and/or the subjunctive in English (cf. section 9.8.3).
(19) If only I were you.
(20) Would that he was here.
(21) I wish that you loved me.
(22) God save the Queen.
(23) May you rot in hell.
CON (= condition): unit which expresses a condition which must be met for
some situation or event to come about. In English, conditions are often
expressed in subclauses initiated by conjunctions like IF and UNLESS, or by
the subjunctive were and/or inversion:
(24) If you trust her, I'll let her in.
(25) Unless you sober up, you will fail the test.
(26) Were this to happen, he would be in serious trouble.
(27) Had I known this, I would have helped her these last few weeks.
The terms used for illocutionary values should not be confused with the
terms used for the form of sentences typically used to express illocutionary
values. Thus a statement often, but not always, takes the form of a declara-
tive sentence and a question often, but not always, takes the form of an
interrogative sentence. Similarly, there is no necessary relationship between
directives and imperative sentences and between exclamations and exclama-
tory sentences.
It is also important to note that the illocutionary values introduced above
do not constitute an exhaustive list. Nor are they meant to be clear-cut
categories allowing a rigid classification of language units. Rather, they are
main types of speech acts into which language units will fit more or less
comfortably. The broad categories of communicative functions listed above
may in fact vary in strength and intensity from language unit to language
unit, and they may blend with each other and with other illocutionary values.
It is useful to think of the speech acts as end points on interacting illo-
cutionary dimensions along which language units may be placed differently
according to illocutionary force. Consider e.g. questions. Questions can be
asked more or less explicitly, with more or less subtlety, according to the
form employed; compare:
(28a) Did John write this book?
(28b) John did not write this book, did he?
(28c) John wrote this book, didn't he?
(28d) John wrote this book, did he?
(28e) John wrote this book?
(28f) I do not know if John wrote this book.
(28g) John wrote this book. ('Actually I do not know, it is my tentative guess,
correct me if I am wrong, am I right so far?')
Though all of these (even (28f,g)) may be construed as questions and, in
appropriate contexts, should be classified as such, the illocutionary force
with which the speaker or writer puts the question varies along the STA-
QUE dimension, with the a-example as the most directly inquisitive, the f-
and g-examples as very nearly pure statements and the others as different
blends of statement and question.
That the illocutionary dimensions are interrelated is also clear from the
following examples:
(29a) Close the window.
(29b) Close the window, will you?
(29c) Do you mind closing the window?
(29d) Could you possibly close the window?
(29e) Gee, it is cold in here!
(29f) According to the building regulation, this window should remain closed at all
times.
(29g) The window is open.
All of these can be construed as directives, but like the questions in (28),
some are more subtle than others. The different illocutionary force can be
described in terms of the different locations of the expressions on the various
dimensions. The two imperatives in (29a,b) are fairly explicit directives. The
a-example is the most unambiguous DIR case of the set. The b-example is
slightly modified by the interrogative tag will you? and thus placed on the
DIR-QUE dimension but fairly close to the DIR end. The yes-no questions in
(29c,d) both express a request beyond a simple yes- or wo-answer and must
be placed on the DIR-QUE dimension, too. The e-example has the form of
an exclamation and expresses a fairly indirect request: it is left to the listener
to conclude that the closing of the window is desired by the speaker. The
example must be placed on the DIR-EXC dimension. The last two examples
belong on the DIR-STA dimension: they have the typical form of statements
but clearly may be spoken in order to make the hearer close the window.
The varying degrees of illocutionary force should not prevent us from
carrying out a rough classification into the main types of speech acts. Thus,
in appropriate contexts, all the examples in (29) should be assigned the
function label DIR. The interesting point that emerges when we review
examples like (29a) to (29g) is that there is a subtle interplay between
function and form. For example, an imperative sentence has more illo-
cutionary force as a DIR than an interrogative or a declarative sentence.
4.4.3. The forms of communicative functions

There are two major points to make in connection with the form of speech
acts. The first point concerns the general form that speech acts may take.
Usually any speech act may take any form: clause, group, compound unit or
single word. The second point concerns the type of sentence involved in
cases where the speech act takes the form of a clause (rather than group,
compound unit or single word). As we have seen, there is no necessary rela-
tionship between speech act and type of sentence (declarative, interrogative,
imperative, etc.). Let us first examine the general form of speech acts.
When we introduced the four main types of speech act (statement,
question, directive and exclamation) in section 4.4.2, we were careful to offer
four examples of each: a clause, a group, a compound unit and a single word.
Here is the full analysis of some of the examples from section 4.4.2 with the
Sent label replaced by the appropriate function:form specification.
(1) James left Brisbane yesterday.
STA
James left Brisbane yesterday

(6) From whom?

QUE
g
H PEP
prep pro
From whom
(11) Smile and be happy!

Dm
cu
CJT CO CJT
v coiy d
Smile and P _C
v adj
I I
be happy
(16) Wow!
EXC
intj
Wow
With regard to the type of sentence involved just in case a speech act is
realized by a clause, it is possible to subspecify the clause label (e.g. 'decl' =
declarative clause, 'inter' = interrogative clause, etc.). On the other hand, such
subspecification of the clause label is largely redundant, just as a subspeci-
fication of groups as noun groups, verb groups, preposition groups, etc. is
largely redundant. The type of clause can in most cases be read off directly
from the syntax or other features of the analysis. Thus, for example, there is
typically subject-operator inversion in interrogative clauses (except when an
interrogative pronoun is subject, as in Who painted this wall?). So unless one
wants to call special attention to the type of clause, there is little motivation
for complicating the clause label.
We want to emphasize the fact that communicative functions signal a link
between sentence and 'larger context' and are hence a subject worthy of
attention in the analysis of larger texts and of utterances in context (as carried
out in 'text linguistics', 'discourse analysis' and 'functional linguistics'). In this
grammar, however, we are primarily interested in the forms adopted by
communicative functions (clauses, groups, compound units, single words)

and their internal structure. Apart from our discussion above showing how a
communicative link can be established in our 'sentence analysis', we shall
have little to say about communicative functions except when they are
relevant for strictly 'sentence-internal', grammatical phenomena.
4.4.4. Non-declarative clauses

As in declarative clauses (cf. section 3.2.4), there are certain regular syntactic
patterns in all the three non-declarative types. In interrogative and exclam-
atory clauses these patterns typically differ from the ones found in
declarative clauses with respect to constituent order, and in imperative
clauses they typically differ from those found in declarative clauses in
lacking subjects.
In interrogative clauses we find exactly the same seven patterns as in
declarative clauses if the subject is realized by an interrogative w/i-word:
SP What's happening?
SPA Who can squeeze through that door?
S PO Who persuaded you?
SPC Who is your source?
SPOO Who will do me this favour?
S ΡΟ C Who would call Quayle clever?
S POA Whatever put that idea into your head?
In wA-interrogative clauses in which the opening wA-word or wh-
construction does not realize the subject there is change of constituent order:
subject-operator inversion, often accompanied by discontinuity. What we
find in such clauses are patterns like A P- S -P (Why did you leave?), O P- S
-P (What do you mean?) and C P S (What good is your assistance?).
In yes-no interrogative clauses there is always inversion:
P- S -P Is your attitude changing?
P- S -P A Can you squeeze through that door?
P- S -P O Must you insult me?
P- S -P C Did they prove uncooperative?
P- S -P O O Could you do me a favour?
P- S -P O C May I call you darling?
P- S -P O A Did you put it in the boot of the car?
In this interrogative subtype there is typically discontinuity as well, as
illustrated by the examples. Note, however, the patterns P S A and P S C in
clauses with BE as the only verb like Is he in Paris? and Is she a doctor?
In imperative clauses, which are typically subjectless, we find the following

six syntactic patterns:
PA Stay here.
ΡΟ Shut the door.
PC Be good.
POO Give me your keys.
POC Knock him unconscious.
ΡΟ A Put it on the shelf.
Note that one-word imperatives like Smile are analysed simply as DIR:v. In
imperative clauses with an overt subject all seven patterns are possible, as in
other clause types, for in addition to patterns with adverbials, complements
and objects we find the pattern S P here: You listen! You behave!, etc.
In exclamatory clauses the following patterns are all possible:
SP What a big crowd turned up!
ASP How beautifully she sings!
OSP What a load of rubbish you're saying!
C SP How delightful it is!
A SPO How confidently you work the machine!
A SPOO How professionally you gave them help!
A SPOC How gently you call me silly!
Some of these patterns are undoubtedly rare - particularly the first one and
the last one - but all of them are grammatically well-formed.
4.4.5. Block language

So far it has been fairly unproblematical to assign a regular form (single
word, g, cu or cl) to complete examples like Spain /An Introduction to
English Sentence Analysis / Jim and Jenny / Pushed to the edge. But there
are sometimes more complex cases of what is sometimes referred to as block
language (headlines, captions, advertisements, notices, telegrams, etc.):
(1) Bad Days at the Bourse.
(2) Alarm bells over nuclear near-misses.
(3) One More Essential for the Overnight Case.
In such examples we usually cannot operate with ellipsis, for there is often
no way of knowing precisely what is missing, if anything (i.e. there is no
unique retrievability, cf. section 4.2). The only thing we can say about (1) to
(3) is that the elements [at the Bourse], [over nuclear near-misses] and [for
the Overnight Case] have adverbial or group-internal dependent function (cf.
Bache et al. 1993: 192ff). If we interpret them as dependents, the whole

utterance can be assigned group status. If, however, we interpret them as
adverbiale, there is no obvious way of determining whether the remainder of
the utterances should be construed as complements, subjects or objects. One
possible solution to this problem is to extend the notion of function stack to
cases where we cannot decide among several established functions, e.g.:
STA
d
A A
g g
Alarm bells over nuclear near-misses
4.5. Vocatives, interjections and dislocation

Simple sentences may be expanded by vocatives. In languages where
syntactic functions are expressed by case forms (e.g. subject: nominative,
object: accusative, indirect object: dative), the term vocative refers to the
case taken by a noun when it is used in the function of an address. This can
be illustrated by a Latin example like Et tu, Brute! 'You too, Brutus!1, in
which the inflectional ending -e signals the vocative.
In English an address is expressed not inflectionally but syntactically,
phonologically and by word class: a proper or common noun or a noun group
is placed initially, medially or finally and is pronounced as a separate tone
unit in the first of these positions and without prosodic prominence as the tail
- or part of the tail - of a tone unit in the last two positions:
(1) Brenda, I could do with a glass of sherry.
(2) We are certainly, sir, treating the death as suspicious.
(3) You must trust me, my darling.
A vocative may also be realized by the personal pronoun you or by a group
containing this word. Except in informal expressions like you guys and you
chaps, this type of address is usually rude:
(4) You (there), I need some service right away.
In those cases where several people are addressed, furthermore, the pronouns
all, everybody, everyone may be used, and if the speaker wishes to catch the
attention of any member of a group, he can use somebody or someone:
(5) Clap your hands, everybody.
(6) Answer the phone, somebody.
Vocatives, interjections and dislocation 107
Vocatives have a number of different subftmctions. In the first place, they

may be used to initiate an act of communication, and when used for this
purpose they are naturally enough placed initially. Secondly, it is common
practice for speakers to ensure that this act of communication is continued by
interspersing their text with references to the listener. Thirdly, vocatives have
various emotive functions. Their use has an implicitly positive effect which
has been described as 'stroking', but they may also be descriptively positive
(darling, dearest John, etc.) or negative (clumsy clot, you idiot, etc.). Finally,
vocatives may signal respect for the addressee and often be required by rules
of etiquette (Your Majesty, Mr President, professor, sir).
A problem which has to be considered is how vocatives should be clas-
sified grammatically. As they behave like sentence adverbiale positionally
and prosodically and have no text-connective function, one possibility is to
classify them as disjuncts. But pragmatically they differ from disjuncts in
being used to express meanings like the ones just mentioned and not to
convey the speaker's comment on the information expressed by the rest of
the sentence or by the style of the expression itself. And formally they differ
from disjuncts in being realized not by adverbs but by (pro)nominals. For
these reasons we shall therefore here consider a vocative a special subtype of
adverbial. Like disjuncts and conjuncts, vocatives have a peripheral status in
the sentence. This can be shown by using the stack convention (cf. section
4.1). For example, (3) can be analysed in the following way:
STA
cl
X
χ g
T
J3_ JP Ο PEP Η
pro g pro pro η
You must trust me my darling
This analysis is supported by the fact that vocatives are prosodically non-
prominent in medial and final position.
In some respects, vocatives are similar to interjections, i.e. to the class of
words which includes emotive words like oh, ugh, mm, etc. and which is here
assumed to include also 'reaction signals' like yes, no, well and greetings like
hi and hello. Traditionally, interjections are defined as words which do not
enter into syntactic relations, and in many cases they do indeed constitute
separate utterances. If the interjection is pronounced with falling intonation
and is separated from the following by a pause, examples like Oh! I see and
Yes! I know must be analysed as a sequence of two utterances with EXC

realized by an interjection followed by STA realized by a clause:
EXC STA
inlj cl
Oh! I see
But interjections may also be pronounced in the same tone unit as the
following clause, and in that case they typically constitute the prehead - or
part of the prehead - of the tone unit, i.e. are pronounced without pitch
prominence and are prosodically connected with what follows. In writing,
this is indicated by a comma: Oh, I see / Yes, I know. As an utterance is
generally assumed to be preceded and followed by silence, it seems
problematic to assume that examples like these constitute a sequence of
utterances. An alternative is to assume that the interjection is a peripheral
adverbial and that the other constituents form a stack:
STA
cl
X.
intj χ
I
Oh, S
pro
I
see
If an interjection does not constitute a separate utterance, it is practically

always placed initially, and if a sentence contains both an initial interjection
and a vocative - as in Yes, sir, that's my baby - the former precedes the
latter. This would seem to indicate that an interjection is more peripheral to
sentence structure than a vocative, and this difference can be captured by
operating with different levels of stacking. An example like Yes, I know, sir -
pronounced as one tone unit in which both the interjection and the vocative
are prosodically non-prominent - can thus be analysed like this:
STA
cl
X
intj χ
Yes X A
χ n
I
I know sir
Vocatives, interjections and dislocation 109
In imperative sentences it is sometimes difficult to separate vocatives from

subjects. The reason for this is that imperative sentences are inherently
vocative, either expressing or, more typically, assuming a second person
singular or plural subject you, representing the listeners) or readers). The
difficulty of distinguishing between imperative subjects and vocatives can be
illustrated by the following examples, which are pronounced as one tone
unit, and in which the initial constituent is prosodically prominent:
(7) Everybody clap your hands.
(8) John and Peter stand over there.
Traditionally, such sentences are held to contain third person subjects, but as
they are synonymous with sentences with final vocatives (Clap your hands,
everybody, etc.), we assume that what we find here are not subjects but
vocatives. These vocatives differ from other initial vocatives in being
pronounced not as a separate tone unit but as an integrated part of a tone unit
spanning the entire sentence. On this interpretation, the subject of imperative
sentences is always implied or explicit you, cf. the presence of the possessive
form your in (7) and the possibility of adding tag questions with you:
(9) Somebody answer the phone, will you?
In informal speech a noun group which does not constitute the comment is
sometimes dislocated to the periphery of a sentence and replaced by a
corresponding pronoun (see Quirk et al. 1985: 141 off). In examples with
ngAi-dislocation, the final noun group is normally pronounced without
prosodic prominence as the tail - or part of the tail - of a tone unit:
(10) She's an excellent pianist, your sister-in-law.
(11) I can't stand him, that friend of yours.
In such sentences - which are characteristic of unplanned discourse - the
noun group tag amplifies the message and serves to ensure that the entity
referred to by the pronoun is correctly identified by the hearer. Left-
dislocation is less common:
(12) Your sister-in-law, she's an excellent pianist.
(13) That friend of yours, I simply can't stand him.
Here the initial noun group is normally pronounced as a separate tone unit
and serves to establish the identity of the entity referred to right away and to
emphasize what constitutes the topic of the sentence. The peripheral
syntactic status of a dislocated noun group - like that of a vocative - can be
captured by means of stacking. For example, (12) can be analysed like this:
STA
Α
g
She's an excellent pianist your sister-in-law
Not only noun groups but also single proper nouns may be dislocated, as
illustrated by He's a decent sort of bloke, Jonathan and Jonathan, he's a
decent sort of bloke. As a dislocated extra identifier is usually pronounced
like a vocative, the function of a proper noun as one or the other usually has
to be determined by the hearer on the basis of the linguistic or situational
context. However, the speaker can make the identifying function unam-
biguously clear by pronouncing a final proper noun as a separate tone unit
(He is a "decent sort of bloke/ "Jonathan) and an initial proper noun as the
stressed beginning of a tone unit spanning the entire sentence ('Jonathan he
is a "decent sort of bloke).
4.6. A final word on advanced sentence analysis

In the preceding sections we have seen a number of construction types which
challenge any simplistic view of language. As the possible analyses of a
particular phenomenon multiply (as they do in cases involving e.g. stacking,
missing constituents and complex predicators), we get a sense not only of a
subjective element of purpose in sentence analysis but also of the complexity
and subtlety of language structure. It is important to realize that an appre-
ciation of language structure may be arrived at by applying a descriptive
apparatus in a consistent manner even if it reflects an incomplete knowledge
of language and even if it seems to impose too rigid a system on the data.
Whenever our descriptive tools fail to cope adequately with data, we may
conclude that we have encountered still another aspect of language that we
do not fully understand yet but which we may come to understand a little
better if we try to incorporate it in our model. It is in this spirit we want our
readers to use our sentence analysis system: as an instrument to approach
language, as a framework for talking about the things we think we
understand and as a means of identifying areas where there are interesting
problems we do not yet know quite how to tackle. There is, of course, often a
practical problem of choosing one out of a number of possible alternative
analyses. Our advice here is that, unless one chooses to focus on a particular
area of interest (say, in order to develop some hypothesis or test some
analytic tool against a broader range of data), one should make one's
sentence analysis as simple as possible.
PART II
5. Constituent order
5.1. Introduction
Constituent order is more rigid in English than in many other languages, e.g.
Italian, Greek, Spanish and Russian. As we saw in section 3.2.4 above, in
declarative sentences with obligatory constituents only, the following
orderings are so common that they may be regarded as constituting the norm:
Structure Examples
SP Nothing happened.
SPA He squeezed between two motor-cars.
S PO You must persuade her.
S P Cs She proved surprisingly uncooperative.
S P Oi Od You do me a very great honour.
S P O Co He had knocked two opponents totally senseless.
S POA We must put some flesh on your bones.
Since these constructions are regarded as the norm, other orderings are
usually described in terms of mobility or movement (such as fronting and
inversion). Consider, for example, the A S P O sequence in:
(1) On the table he put the book.
Here the adverbial is regarded as having moved from the designated A
position of the regular S P O A structure to a position before the subject for
reasons of style or focus, for example.
The picture of English sentence structure is, however, not complete unless
we consider the role also of optional adverbials. In our discussion of
adverbials in section 3.2.9, we saw that there were very few general
restrictions on the position of optional adverbials and that an adverbial is
often equally happy in a number of different positions in a sentence. In
declarative sentences, the adverbial is in fact the most mobile constituent.
By contrast, the other clause functions (subject, predicator, objects and
complements) are more firmly fixed in their designated positions in the
structures listed above. But principled deviation does occur. Like adverbials,
for example, objects and complements may, under certain conditions, be
placed in sentence-initial position. In reply to a question like What do you
see?, for example, a speaker may say That I -will tell you another time, i.e. he
may place the direct object initially (Od S P Oi A). Fronting of a complement
112 Constituent order
can be illustrated by an example like Waddle his name is (C S P). Another

important deviation from the norm is the relative positions of S and P in
sentences like Here comes Bill and Did you see her again?
When constituent order is compared across languages, what is compared is
the 'favourite* order found in each language. While the favourite order of pre-
dicator, subject and object is S P O in a large number of languages including
English, it is S O P in many other languages (e.g. Turkish, Japanese), and hi
a third group of languages it is P S O (e.g. Arabic, Hebrew).
When constituent order is discussed within a particular language, such as
English, it is done so in terms of the norm identified in declarative sentences
and deviation from this norm. Central questions which have to be considered
are 'What does constituent order signal communicatively?' and 'What are the
conditions under which the normal constituent order is changed?'
5.2. Functions of constituent order

5.2.1. Grammatical relations
Constituent order performs a number of different functions. One is to signal
grammatical relations. Consider the following example (which refers to the
historical situation right after the Napoleonic wars):
(1) England gave Sweden Norway. (S P Oi Od)
Here the subject, object and indirect object are identified through constituent
order exclusively. When confronted with it, the hearer will therefore
understand England to be the agent, Sweden the beneficiary and Norway the
affected country. If we change the order of the three nouns while retaining
the predicator in the second position, the locations of the subject, object and
indirect object will remain unaffected. The subject will still be in the first
position, the direct object in the last position and the indirect object in the
last but one. This implies that reversal of the last two words, for example,
will be taken by the hearer to mean that it was Sweden which was given to
Norway. One among several interacting principles of constituent order is that
of functional stability: constituents with the same syntactic function tend to
be placed in the same position (cf. Dik 1989: 343). We can therefore speak of
a characteristic subject position, a characteristic object position and so on.
5.2.2. Illocutionary value

Another function of constituent order is to give information about the
illocutionary value of an utterance, i.e. about the speech act it is used to
perform. As we saw in section 4.4.2, four basic speech acts can be
Functions of constituent order 113
distinguished: statement, question, directive and exclamation. Recall that,

roughly speaking, statements give information to the listener, questions seek
information from the listener, directives instruct the listener to perform an
action and exclamations tell the listener that the speaker reacts emotionally
with surprise, disapproval, pleasure, etc. In section 4.4.4, these basic illo-
cutions were found to be relevant to the question of constituent order. All
languages appear to have special sentence structures associated with
statements, questions and directives (declarative, interrogative and imper-
ative sentences respectively), and many languages have a special sentence
structure for exclamations as well (exclamatory sentences). Although there is
a far from necessary relationship between illocutionary value and sentence
form, the association between form and function is strong enough for the
form to serve as a general signal of the function. In other words, when we
use a declarative sentence it can in general be interpreted as a signal of the
illocutionary value of statement; and so forth.
If we restrict our attention to transitive sentences, the difference between
the sentence types can be symbolized and illustrated like this:
declarative: SP O You must learn grammar.
yes-no interrogative: P- S -P O Must you learn grammar?
wA-interrogative: O P- S -P What must you learn?
SPO Who must learn grammar?
exclamatory: O SP What you must learn!
imperative: ΡΟ Learn grammar.
In very general terms we can say that the constituent order P- S -P will guide
the listener towards the illocutionary value 'question1. The order O S P where
Ο is realized by What will guide the listener towards 'exclamation', and Ρ Ο
alone will make him expect the illocutionary value to be 'directive'. S P O is
the typical order found in declarative sentences, but it is also found in
imperative sentences with a subject and in wA-interrogative sentences in
which the subject contains the interrogative word. Though most likely to
signal 'statement', this constituent order is therefore a more unreliable guide
to illocutionary value than the others.
5.2.3. Information structure

A third function of constituent order is to signal information structure, i.e.
the way in which a message is organized into information units. The organ-
ization of a message often reflects a division between given information, i.e.
what is assumed to be known to the hearer, and new information. Normally
the speaker will proceed from what he assumes to be known (the topic or
theme) to what he assumes to be new (the comment or rheme). This can be

illustrated by an example like The letter on the desk is from your mother, in
which the information contained in the first part (the subject noun group The
letter on the desk) is assumed and that contained in the second part (the
predicate stack is from your mother) is new and at the centre of the speaker's
communicative interest. The general tendency towards placing new and
particularly important information at the end of the sentence is called the
principle of end-focus.
The use of passive vs. active voice in English is largely determined by the
way the speaker wishes to organize his message in terms of given and new
information. In an active example like Arsenal beat Tottenham, the speaker's
point of departure is Arsenal while the rest of the sentence is at the centre of
his communicative interest. In the passive equivalent Tottenham were beaten
by Arsenal, on the other hand, it is Tottenham which constitutes the point of
departure. Though the basic meaning is the same in these two sentences, the
presentation is different in terms of given, new and focus. In this respect the
passive sentence is similar to an active sentence like Tottenham lost to
Arsenal. In passive sentences the initial constituent tends not only to express
given information but also to have a connective function in the text:
(1) This was on a rotation of six days a week at the restaurant with one off
(Thursday). That day was spent in recuperation on his back on the sofa,
generally with open eyes, for his feet ached after the hours of standing.
Here the subject noun group That day refers back to the day off referred to in
the preceding sentence.
In order to highlight part of a message, the speaker may place a sentence
constituent in a special position, i.e. he may for this purpose deviate from the
typical orderings described above. For example, he may place an object or a
complement in the initial position of a declarative sentence. Such placement
of a constituent in marked first position is cv\\ea fronting:
(2) That story I will tell you another time. (Öd S P Oi A)
(3) Chris Waddle his name is. (C S P)
In moving an object or complement from its usual position to the initial
position, the speaker gives prominence to this part of the message, i.e. he
instructs the listener to pay special attention to it. Highlighting by means of
fronting is particularly common if the speaker can in this way simultaneously
establish direct linkage with the preceding part of a conversation or of his
own message. This illustrates that the position of a constituent is dependent
not only on factors within the sentence in which it occurs but also largely on
the relation between that constituent and the preceding text.
Functions of constituent order 115
(4) Speaker A: What do you see?

Speaker B: That I will tell you another time.
While fronting of objects and complements is relatively rare in declarative
sentences, adverbials are frequently placed in initial position to give prom-
inence to them and to link them to the preceding text:
(5) Lo' s room contained his iron bed and one chair, over which he had thrown
his black trousers and a shirt. In a corner of the room was a gas ring, a kettle
and a single glass. (A P S)
Part of a message may be highlighted not only through fronting but also by
means of postponement. One example of this is extraposition of the subject
(cf. section 3.2.2):
(6) It worries me that the children have not returned. (Sp P O Sr)
(7) It would be awkward to go now. (Sp Ρ C Sr)
Not only subjects but also objects may be postponed to give them special
prominence:
(8) The court pronounced not guilty a black man. (S P Co O)
(9) The court pronounced not guilty the woman who had been charged with the
murder. (S P Co O)
In English as in other languages there is a preference for constituents to
occur in an order of increasing heaviness. This is called the principle of
increasing complexity, or of end-weight.
Object postponement may also involve extraposition (cf. section 3.2.5):
(10) I find it a challenge writing this report. (S P Op Co Or)
Informationally, the type of highlighting obtained by postponement is not the
same as that obtained by fronting. By means of fronting the speaker informs
the hearer that the element thus moved constitutes the topic (or theme), i.e.
the entity about which something is said (see section 3.2.2). By means of
postponement, on the other hand, he informs the hearer that the element thus
moved is at the focus of his communicative interest. For these reasons
highlighting by means of fronting is termed topicalization (or thematization)
while highlighting by means of postponement is termed focalization.
To recapitulate: constituent order is used to signal syntactic functions
(subject, object, etc.), to give information about the speech act an utterance
performs (statement, question, directive, exclamation) and to signal how a
message is structured informationally (what is highlighted through topical-
ization or focalization).
5.3. Inversion
5.3.1. Preliminaries
The term inversion (from Latin inversio/inversionis 'turning around') is used
in grammar to refer to a reversal of syntactic constituents. This may be
illustrated by an interrogative sentence like Was it disappointing?, in which
the predicator precedes the subject. The order P S may here be seen as a
result of a syntactic change which reverses the order S P found in a
declarative sentence like It was disappointing. Inversion in its broad sense, it
should be pointed out, refers not only to reversal of subject and predicator (or
operator). In The court pronounced not guilty the woman who had been
charged with the murder, the order Co Ο may be seen as a reversal of the
normal order Ο Co found in an example like The court pronounced the
woman not guilty. In She placed on the topmost shelf her entire collection of
grammar books, similarly, the order A O may be seen as a reversal of the
normal order Ο A found in She placed her books on the topmost shelf.
Furthermore, inversion operates not only on the level of sentence structure
but also on the level of group structure. For example, the ordering of an
example like He had no patience with problems hypothetical may be
regarded as a reversal of the ordering in He had no patience with hypo-
thetical problems. In descriptions of English grammar, however, the term
inversion usually refers to reversal of subject and (part of the) predicator, to
which we now turn.
5.3.2. Full and partial subject-predicator inversion

Depending on whether it is the entire predicator or only part of it (the oper-
ator) which changes places with the subject we distinguish between y«//
inversion and partial inversion. Full inversion, which is often optional, can
be illustrated by the following example with the sentence structure A P S :
(1) In rhyme and rhythm resides a certain magic power.
Partial inversion, which is usually obligatory, is found in the next example,
which has the sentence structure A P- S -P O A:
(2) Only with difficulty had Lily explained her conduct to her parents.
As is apparent, only the first of the two verbs realizing the predicator - the
operator - is here inverted with the subject. If there are more verbs than two
in a predicator group, it is the first of these only (the operator) which changes
places with the subject. For this reason partial inversion is often referred to
as 'subject-operator inversion':
Inversion 117
(3) Only with difficulty could Lily have explained her conduct to her parents.
In those cases where a sentence without inversion has a predicator which is
realized by a single verb, the corresponding sentence with partial inversion
requires DO-support (cf. section 3.2.1):
(4) Only after a while did he notice that his mother was crying.
As will be recalled from section 3.2.1, special rules apply to the primary
verbs BE and HAVE: sometimes they behave like full verbs, sometimes like
operators. If the predicator is realized by the verb BE there is in interrogative
sentences no DO-support. For example, the sentence with inversion which
corresponds to He is comfortable is not *Does he be comfortable? but Is he
comfortable? What we find here therefore looks like full inversion in that it
is the entire predicator which is moved. Nevertheless, we shall treat reversal
of a subject and a form of BE under partial inversion in those cases where
subject-operator inversion occurs generally. We shall thus treat an example
like Are you comfortable? together with Have you been comfortable? and
Never was I so deeply in love together with Never have I been so deeply in
love. On the other hand, the reversal of a subject and a form of BE found in
examples like Here is the milkman? and On the walls were pictures of half-
naked women and colourful landscapes will be dealt with under full
inversion, i.e. together with Here comes the milkman and On the walls hung
pictures of half-naked women and colourful landscapes, because it occurs
under those conditions where this subtype of inversion occurs generally.
In BrE, there is not invariably DO-support in inverted sentences if the
predicator is realized by the verb HAVE. This can be illustrated by an
example like Have you any doubt about his guilt? As in the case of sentences
with BE, for example Are you in doubt about his guilt?, we shall treat this
type of reversal under partial, subject-operator inversion, in spite of the fact
that it is the entire predicator which is moved. The reason for this is once
again that reversal of a subject and a form of HAVE is found under those
conditions where this obligatory subtype of inversion occurs generally. We
shall accordingly treat examples of the type Have you any doubt about his
guilt? together with the less formal and more common varieties Do you have
any doubt about his guilt? and Have you got any doubt about his guilt?
Partial inversion is used under two well-defined conditions. In the first
place it serves to signal illocutionary value, most importantly to perform the
speech act of posing a question. Secondly, it is used in those cases where a
sentence contains a special opening constituent, most importantly a negative
or restrictive opener.
5.3.3. Partial inversion with illocutionary value

The most important role played by partial inversion in English is to pose a
question by way of forming an interrogative sentence. As illustrated by the
following examples, it is by this means that a speaker typically elicits inform-
ation from the listener about ayes-no relationship:
(1) Could you live without me?
(2) Have you liked me for a long time?
(3) Did they let him go?
(4) Are you sure I can't fix you a sandwich?
Yes-no questions, it should be added, can also have straight constituent order
and be signalled by other means. In spoken English this normally requires
that they are uttered with rising intonation. The following examples are thus
usually pronounced with a final upward tonal contour:
(5) So you never had a chance to talk to him again?
(6) And you made up your mind to come and see me first?
In English (unlike spoken French, for example) partial inversion is the
general rule in yes-no questions, however. It is found also in tag questions,
i.e. in questions where a construction with inversion of the type isn't it, has
he, usually consisting of an operator and a personal pronoun, is added to a
non-interrogative sentence:
(7) This is the tendency, isn't it?
(8) You didn't feel very well, did you?
Partial inversion is also found in wA-interrogative clauses and here typically
performs the function of posing a wh-question. In contrast to yes-no
questions, however, wA-questions are signalled as such not only by inversion
but also by a special question word. If this realizes the subject or part of the
subject, there is straight subject-operator order (e.g. What happened?) and
the illocutionary value is signalled by the wA-word exclusively. On the other
hand, wA-words occur also in statements (for example 777 say what I mean)
and in exclamations, so they do not in themselves signal the speech act of
posing a question either. In examples like the following, the difference
between question and exclamation is signalled exclusively by the ordering of
subject and operator:
(9a) How can you say that?
(9b) How you can say that!
(lOa) What sounds can you make?
(lOb) What sounds you can make!
Inversion 119
Partial inversion in wA-interrogative clauses can be illustrated by the

following examples:
(11) What do you mean, the police?
(12) Why are you always making such a fuss about your brother?
(13) How could you go to bed with somebody you didn't love?
(14) Which way should we go?
The optative speech act of expressing a wish - or a malediction - can in
English be performed in several ways, some of which do not involve
inversion. For example, a speaker may for this purpose use the performative
verb WISH (after the personal pronoun / functioning as subject) or the
performative opener If only. This can be illustrated by / wish / If only the
performance would begin soon. He may also, though this strategy is much
rarer, express a wish by means of the subjunctive mood (cf. section 9.8).
Wishes signalled in this way are frozen expressions, of which some have
straight constituent order and others take inversion:
(15) God save the Queen.
(16) Heaven forbid.
(17) Long live Trotsky.
(18) Suffice it to say that your essay is unsatisfactory.
(19) So be it.
Wishes and maledictions, however, can also be expressed by sentences in
which the verb MAY occurs, and in this case there is partial inversion:
(20) Please may it have been instantly.
(21) May she rot in hell.
That inversion is crucial here appears from the fact that utterances withMAY
in which the order of constituents is straight, e.g. She may rot in hell, are not
used to express a wish but a statement.
As the use of utterances with inversion and MAY for the expression of
wishes is rare, this particular type of inversion with illocutionary value is of
rather minor importance.
Finally, to anticipate section 5.4 on constituent order in subclauses, there
may be partial inversion in the subclause of a complex sentence which
indicates conditional commitment on the part of the speaker. In English,
conditional sentences are normally signalled by if or unless and less
commonly by words or word groups such as supposing, in case and
assuming (that). In formal style, however, they can also be signalled by
partial inversion (with had, should or were as operator):
(22) Had I known this, I would never have accepted the offer.
(23) Should he do that, I will crucify him.

(24) The total output would be much worse were it not for the winter crop.
What the speaker indicates by inversion here is that the subclause performs
the function of laying down a condition, typically hypothetical, on which the
state of affairs described in the matrix clause is dependent.
Before leaving partial inversion with illocutionary value, we should
mention that this constituent order is sometimes also found in directives and
exclamations. As illustrated by Don't you let him force you back, this is the
constituent order we find in negative imperative sentences with an expressed
subject. And as illustrated by Isn't she lovely! (pronounced with falling
intonation) there is, of course, inversion in yes-no interrogative sentences not
only if they function as questions but also as exclamations.
5.3.4. Partial inversion caused by an initial constituent

There is partial inversion in sentences beginning with a negative or
restrictive constituent other than the subject. This constituent may often be
regarded as fronted and is in most cases an adverbial:
(1) Nowhere are the effects of these policies more evident than in Denmark.
(2) Under no circumstances must you leave the room.
(3) Rarely have I set eyes on such a stunning beauty.
(4) Not another pound will you get from me. (fronted negative object)
(5) Only after a while did he notice that his mother was crying.
(6) Hardly a word did he utter, (fronted restrictive object)
If a negative word does not apply to the entire sentence but only locally to
the constituent in which it occurs, there is straight subject-predicator order:
(7) Not long ago my grandmother turned eighty.
In examples of this type with local negation the sentence as a whole
expresses a positive statement (note that Not long ago could be replaced by
Recently). In such cases, where the event described by the part of the
sentence following the negative opener is true irrespective of this opener,
partial inversion is not used. Attention should also be drawn to examples
without inversion of the following type:
(8a) Only last month we could cope with the increase in demand.
Here the initial adverbial is not really restrictive but has a temporal meaning
which can be captured by the paraphrase 'as late as last month'. If the word
group only last month does have restrictive meaning, i.e. if it means 'last
month and no others', it will trigger inversion:
Inversion 121
(8b) Only last month could we cope with the increase in demand.
In spoken English, partial inversion after a negative opener is particularly
common in clauses beginning with one of the pro-forms NEITHER and NOR
(cf. section 4.2.3). Examples of inversion after NEITHER:
(9) I didn't turn up and neither did my wife.
(10) Speaker A: I can't swim. Speaker B: Neither can I.
If we compare these with / didn't turn up and my wife didn't turn up either
and / can't swim either, it seems reasonable to assume that n't... either is
fronted in the contracted shape of neither to underline its connective function
and that the main verb is dropped because it is redundant in the context.
Inversion in clauses with NOR can be exemplified by a dialogue like
(11) Speaker A: I don't like her. Speaker B: Nor do I.
Partial inversion after NOR in a full sentence is characteristic of formal style:
(12) I have never procrastinated. Nor do I intend to start doing so now.
Initial placement of a negative or restrictive constituent clearly involves
information structure, for it is motivated by a wish to topicalize this part of
the message. For example, a sentence like Never have I heard such nonsense
differs from / have never heard such nonsense in that it topicalizes a con-
stituent with negative meaning through placement in the initial position.
What is created by this constituent order is often a double focus: the negative
constituent is highlighted through placement in the initial position (topical-
ization of Never) and another constituent is highlighted by occupying the
final position (focalization of nonsense), which according to the principle of
end-focus is typically reserved for new information.
While initial placement of a negative or restrictive constituent relates to
information structure, the inversion which accompanies it cannot plausibly
be explained in this way as well. A double focus could just as well be created
with straight order as with inversion, for example by the order in an
ungrammatical sentence like *Never I have heard such nonsense. As the
initial constituent pulls the first part of the predicate into the position before
the subject, this type of subject-predicator reversal may be characterized as
attraction inversion (see Hartvigson & Kvistgaard Jakobsen 1974: 25) His-
torically it reflects a stage of the English language when inversion after all
kinds of initial constituents other than the subject was the general rule.
We now turn to partial inversion in sentences beginning with the adverb
SO or SUCH. In examples of the following type, SO/SUCH is part of a dis-
continuous complex constituent - object, complement or adverbial - which
ends in a correlative /Aa/-clause (i.e. a clause regularly accompanying

SO/SUCH but separated from it by other linguistic material):
(13) So much did he eat that he was almost sick. (O- P- S -P -O)
(14) Such was the heat that she was unable to finish the recital. (C- P S -C)
Through partial fronting, the premodifier So and the head much in (13) have
been separated from the postmodifying fAaf-clause. In (14) it is the
pronominal head Such that has been separated from its dependent clause
through partial fronting. In both examples the effect is one of double focus. If
no fronting takes place, as in He ate so much that he was almost sick, the
complex constituent introduced by SO/SUCH only has end-focus. But in the
partially fronted variations of such sentences both the postmodifying thai-
clause and the head word of the fronted constituent are highlighted.
In spoken English, partial inversion triggered by the pro-form SO is par-
ticularly frequent in clauses of the same type as those mentioned above
beginning with NEITHER or NOR (cf. also section 4.2.3):
(15) I turned up and so did my wife.
(16) Speaker A: I can swim. Speaker B: So can I.
Such sentences may be compared with sentences without inversion and pro-
forms like / turned up and my wife turned up too and / can swim too. Note
that the adverb so is not here a marker of degree but means 'also1.
In formal style, attraction inversion is also found in comparative sub-
clauses with the subordinator conjunction as or than:
(17) Crabs were fresh and plentiful in the market at the moment, white vegetable
was good, as were Holland beans.
(18) Alice knew Peter far better than did most of her classmates.
In such constructions, the operator serves as a pro-form representing a full
predicate (P C or Ρ Ο).
Before leaving partial inversion caused by an initial constituent, we should
point out that it is sometimes also found after other openers than the ones
mentioned above, for example WELL in the sense of'fully' and RATHER:
(19) Well do I know that you hate me.
(20) Much rather would I be dead.
As a manner adjunct THUS is particularly versatile as an opener: it allows
both partial and full inversion with little or no difference of meaning:
(21) Thus did a lovely evening end. / Thus ended a lovely evening.
Inversion 123
5.3.5. Full inversion and information structure

While partial inversion is used to signal illocutionary value, or is mechan-
ically triggered once the speaker (for reasons of information structure) has
chosen to place a negative or restrictive constituent initially, full inversion is
a matter of information structure. As pointed out in section 5.2.3, a speaker
may move a constituent to a special position to give prominence to it. If he
moves an object, complement or adverbial to initial position, he may in this
way establish linkage with the preceding part of the text, i.e. such a con-
stituent is not only highlighted through fronting (i.e. topicalized) but also
acquires a connective function. When such fronting is called for in S P O, S
P C and S P A sentences, the predicator will come to occupy the final
position - and thus get end-focus - if it is notfrontedas well by being moved
to the position before the subject. As illustrated by That story I don't believe,
end-focus on the predicator is by no means ruled out in sentences with
topicalization. But in sentences with fronted adverbials of certain types -
particularly adverbials of place - it tends to be avoided. This can be
illustrated by an example like In the middle of the room stood a table (A P
S), in which the predicator and the subject normally cannot change places.
While inversion in connection with fronting of an adverbial appears to be
due primarily to a tendency to avoid end-focus on the predicator, it may be
supported by the principle of end-weight. For example, this principle would
be badly violated if the fronting of an adverbial in a sentence like A vat of
boiling water stood in the middle of the room were to be unaccompanied by
subject-predicator inversion. If the structure of this sentence is changed from
S P A not just to A S P but to A P S, however, the resulting sentence is as
well balanced as the one with normal S P A order: In the middle of the room
stood a vat of boiling water has just as much end-weight as A vat of boiling
water stood in the middle of the room.
If a place adverbial can be fronted without resulting in final placement of
the predicator, and without throwing the sentence off balance, inversion does
not take place, or is not obligatory:
(1) In the distance a sunlit range of mountains could be seen very clearly.
(2) On the doorstep women sat nursing their babies and gossiping.
(On the doorstep sat women nursing their babies and gossiping.)
Generally speaking, fronting of adverbials (and of complements and objects
as well) is usually ««accompanied by inversion, and in this respect English
differs from e.g. German, Dutch and the Scandinavian languages. This is
illustrated by the next examples, in which inversion would be impossible:
(3) After the birth of their son, Lily had been unable to cope with all the
housework.
(4) On this occasion the young engineer capitulated without a struggle.

In the few cases where complements and objects are fronted, full inversion is
decidedly rare. After fronted adverbials, however, it is not uncommon, par-
ticularly in literary language.
5.3.6. Full inversion after a fronted adverbial

The following examples illustrate inversion accompanying fronting of an
obligatory adverbial:
(1) On the walls were pictures of half-naked women and colourful landscapes.
(2) At his side sat a black Alsatian dog.
(3) On street corners appeared posters with long lists of executed persons.
(4) In rhyme and rhythm resides a certain magic power.
In these examples fronting serves the purpose of establishing narrative
continuity. In all of them, it will be observed, fronting unaccompanied by
inversion would result in structures which to a higher or lower degree - and
in most cases flagrantly - would violate the principle of end-weight. As
illustrated by the examples, it is typically fronting of adverbials of place
which is accompanied by inversion. Fronting of adverbials of time, however,
may also necessitate inversion, for instance in Now comes the time we've
been waiting for so long. The verbs occurring in sentences with fronted
adverbials and inversion are typically BE and other intransitives such as
APPEAR, HANG, LIE, SIT and STAND. Potentially transitive verbs (such as
SEE) are only possible in such sentences if the construction as a whole is
intransitive. This can be illustrated by a passive example like In the distance
could be seen a sunlit range of mountains, which has the intransitive
structure A P S . This example also demonstrates that the predicator which
changes places with the subject may be realized by a complex verb group.
In ordinary, informal speech, full inversion after a fronted adverbial is not
very common. We find it in short intransitive sentences beginning with one
of the adverbs of place HERE and THERE:
(5) Here comes the bus.
(6) There goes the last bus.
There may also be inversion in intransitive sentences beginning with one of
the directional adverbs DOWN, UP, IN and OUT. Such a fronted adverb may
realize an independent adverbial:
(7) Out darted a mouse.
However, it may also realize the second part of a phrasal predicator:
Inversion 125
(8) Up went the prices once again.

As appears from the examples, the verbs occurring after HERE, THERE and
directional adverbs in sentences with inversion typically express movement:
COME, DART, FALL, FLY, GO, JUMP, etc.
The common factor in all the examples with a fronted adverbial and
inversion is intransitiveness. While fronting of adverbials accompanied by
inversion in literary language usually serves the purpose of establishing
narrative continuity, fronting in examples like the ones just given from
informal spoken language does not usually have this specific function but
creates a more vivid impression.
5.3.7. Full inversion after a fronted participial predication stack

A special form of inversion - related to the one found after fronted
adverbials - is found in sentences where a present or past participle is
moved from a complex predicator to the initial position together with an
adverbial (or with more adverbials than one) which accompanies it. Such
fronting of predication stacks is characteristic of literary language. Instead of
writing The Curry Mahal was sandwiched between the Chinese restaurants,
for example, an author may write Sandwiched between the Chinese
restaurants was the Curry Mahal. Here are some more examples:
(1) Standing around in the shops were heavily bearded men in long gaberdine
coats, wearing thick boots.
(2) Heaped on a lectern near the window were religious books and sheets of
paper covered with writing.
In sentences of this type, not only has the order of the subject and the
predicator been reversed but also that of the two constituents in the complex
predicator. The fronted adverbial and verb typically indicate location and
posture respectively. In moving participle + adverbial(s) to the front of the
sentence, the writer gives prominence to this part of the message, particularly
to the state described by the initial verb. If he fronts only the locative
adverbial(s), it is only this part of the message which is highlighted. For
example, the meaning of heaped is not emphasized in an example like:
(3) On a lectern near the window were heaped religious books and sheets of
paper covered with writing.
It should be pointed out that predicators which potentially describe dynamic
situations such as HEAP, LAY, etc. describe Stative situations exclusively if
the participle is fronted. Out of context an example like Religious books were
heaped on a lectern near the window is potentially ambiguous in that the
situation it describes can be either a stable state of affairs or involve a
change. In Heaped on a lectern near the window were religious books, on the
other hand, the situation referred to can only be understood to be Stative. In
addition to having a highlighting function, participle + adverbial fronting
may thus be chosen to avoid ambiguity.
Fronting of a present participle in sentences with full inversion is found
also in cases where it is not an adverbial but an object which accompanies it:
(4) Awaiting them were a tray of sandwiches, two bottles of wine, the director in
uniform, and, to top it all off, an exceptionally beautiful girl.
(5) Filling the chamber was a grand plan of the City of Willows, metaphor of
that legendary matrix of Hung heroes.
In examples of this type, the subject is heavy and the rest of the sentence
light. One way in which the speaker can observe the principle of end-weight
is therefore to front part of the predicator together with the object.
Sentences with fronted participles of the type exemplified above often vary
with less 'literary' sentences in which there is used as provisional subject
(and in which there is therefore not inversion):
(6) Heaped on a lectern near the window there were religious books and sheets
of paper covered with writing.
As this 'existential there' is normally used to introduce new entities into the
discourse (cf. section 3.2.2), the possibility of using it in sentences with
fronted participles demonstrates that the information expressed by the heavy
structure realizing the subject is new. On the other hand, the elements fronted
here do not necessarily specify given information or have the cohesive
function that initial constituents often have. What they typically do is set the
scene for the state of affairs described by the rest of the sentence.
Sentences with inversion and fronted participles are also common in news
reporting. Here the fronted constituents usually do have cohesive function
and refer to given information:
(7) Also killed in the shoot-out were three teenagers from the Bronx.
The analysis of a fronted participle as part of the predicator is not the only
one which presents itself. The participle and the construction which is
fronted together with it might instead be analysed as an adverbial (in the case
of a present participle construction) or as a complement (in the case of a past
participle construction). In support of such an alternative analysis it may be
argued that sentences like the following are possible, in which the finite verb
is not a form of BE:
(8) Awaiting them stood a group of heavily armed soldiers. (A P S)
(9) Heaped on a lectern near the window lay religious books and sheets of paper
covered with writing. (C P S)
Inversion 127
As such sentences are syntactically closely related to sentences in which the

finite verb is a form of BE, and as the connection between BE and a fronted
participle appears to be looser than that between BE and a following
participle, the analysis in sentences with BE of a fronted participle construc-
tion as an adverbial or a complement cannot be ruled out as an alternative.
5.3.8. Full inversion after a fronted subject complement

In the relatively few cases where a subject complement is fronted in English,
the ordering of subject and predicator may be straight or inverted. The
former has already been illustrated by the example Waddle his name is. If a
subject is heavy, however, inversion is possible in order for the principle of
end-weight to be observed:
(1) Strange indeed was her urge to tell everybody what she thought of them.
Inversion after a fronted complement is chiefly found in clauses with BE, and
where the complement refers to given information and is realized by a
comparative construction which refers to a preceding part of the text:
(2) A course in trigonometry may be useful enough, but far more useful would
be a refresher course in elementary maths.
(3) Grammar is undoubtedly relevant, but equally relevant are recent approaches
to language learning.
In selecting the constituent order C P S in the second clause of such
compound sentences, the speaker accomplishes three things: he gives
prominence to the part of the message covered by the group realizing the
complement, he establishes a direct link to the preceding clause and he
observes the principle of end-weight.
5.3.9. Full inversion after a fronted object

Objects are rarely placed initially in English. As pointed out in section 5.2.2,
this placement is found in wA-interrogatives of the type What must you
learn? where the interrogative pronoun realizes the object. And as mentioned
in section 5.2.3, an object may be fronted in a declarative sentence if the
speaker wishes to highlight the part of the message its realization spans: That
story I will tell you another time. Such fronting, which is unaccompanied by
inversion, usually also has a cohesive function, as in the following example,
where the fronted object pronoun This refers back to the preceding clause:
(1) He knew what he liked and Lily didn't conform to the specifications. This he
knew with a certainty as absolute as his knowledge that the food he served
from the 'tourist1 menu was rubbish.
Fronting of objects accompanied by inversion is chiefly limited to reporting

clauses, in which a verb of saying in the simple present or past tense takes
the clause quoted as its direct object (O P S):
(2) 'Eh?' said the driver.
(3) didn't hear anything,' answered the woman's voice.
In examples like these, inversion, though not obligatory, is chosen in order to
observe the principle of end-weight. If a reporting clause contains an
adverbial, this principle can be observed without changing the normal
ordering of subject and predicator (O S P A):
(4) 'We are going to visit a different town,' their driver announced cheerfully.
The order of subject and predicator is also straight if the reporting clause
contains an indirect object (Öd S P Oi):
(5) 'That's your father in the car, Son,' Lily informed him.
If the subject of a reporting clause is a personal pronoun, which is even
lighter than a verb of saying, there is naturally enough no inversion:
(6) 'This is madness,' he said.
In earlier stages of English, inversion of pronoun and verb of saying was
common, and thus has an archaic, poetic or even comical ring to it:
(7) "Tis madness,' quoth he.
This order is very occasionally used in modem English in narrative contexts
with the predicator and pronominal subject of the reporting clause in pre-
object position ('initial reporting clauses'):
(8) Says he: 'Sit down. My God, you are glistening like an apple!' Says she:
'Well, I'd rather not, you see, I'm fed up with your condescending manners.1
The effect here is to emphasize the conversational turn-taking, with a more
clearly felt contrast between the interlocutors.
In everyday conversation, inversion is not uncommon in cases like:
(9) Speaker A: 'Jack can do it.'
Speaker B: 'Says who!1
(10) Speaker A: 'John's wife can run a mile in four minutes.1
Speaker B: 'Yeah, says he!'
In both cases the pronoun is fully stressed, and the construction conveys a
tone of disbelief or contempt.
Finally, both inversion and straight order are used in reporting clauses whose
subject is realized by a proper noun and in which there is no adverbial or
indirect object:
Inversion 129
(11) Ί am no coward!' declared Xavier / Xavier declared.

If there is no object fronting, i.e. if the reporting clause is placed initially, the
ordering of subject and predicator is straight:
(12) Then Jane said, Oon't be angry with me, I beg you!' (A S P O)
In journalism, however, there is occasionally inversion in initial reporting
clauses, in accordance with the principle of end-weight:
(13) Explained Clay Mulford, the national group's general counsel: *We don't
want to be held responsible for every action they take.' (P S O)
The conditions under which subject and predicator are inverted in English
are summarized in table 1.
question Have you finished?
illocutionary wish May you have a happy married life!
conditional commitment
partial Had I known this, I would
(subject- never have accepted the offer
operator)
initial negative/restrictive constituent

attraction Never have I heard such nonsense!
Subject- initial so/such/as/than constituent

predicator So much did he eat that he was
inversion almost sick
fronted adverbial
Here comes the bride
fronted complement
full information | Far better would be a famine

relief operation
structure
fronted object
"Watch out!" said the driver
fronted participial predication stack

Standing around in the shops
were heavily bearded men in long
Table 1: Subject-predicator inversion gaberdine coats, wearing thick boots
5.3.10. Inversion of other sentence constituents

As pointed out in section 5.3.1, inversion may be understood to refer not
only to reversal of subject and predicator but also to reversal of other
sentence constituents. In nearly all cases such re-orderings can be explained
as caused by the principle of end-weight.
Inversion of an adverbial and an object or complement can be illustrated by
the following examples (the first of which is repeated from above):
(1) I just saw on television how some Indian people started a shop and put the
old grocery on the corner out of business. (S A P A O)
(2) When squeezed into regular metres, an amorphous world becomes at once
orderly, lucid, clear and beautiful. (A S P A C)
These sentences should be compared with sentences with the normal orders S
Ρ Ο A and S P C A like I just saw a show on television and (When squeezed
into regular metres,) an amorphous world becomes orderly at once.
Inversion of an object and an object complement is illustrated by the next
example (also repeated from above):
(3) The court pronounced not guilty the woman who had been charged with the
murder. (S P Co O)
This sentence may be contrasted with The court pronounced the woman not
guilty, which has the normal constituent order S P O Co.
In sentences with two objects, the indirect object virtually always comes
first (according to the principle of functional stability, cf. section 5.2.1):
(4) The remittances gave the couple a comfortable existence. (S P Oi Od)
(5) Give me the whisky bottle. (P Oi Od)
For exceptions like Give it me (P Od Oi), see section 3.2.6.
While the order Oi Od is thus relatively fixed, considerations of weight
may lead to a reversal of the order plus recategorization of the indirect object
as an adverbial preposition group (cf. section 3.2.6):
(6) I gave the old man my umbrella.
(7) I gave my umbrella to the old man with the big black labrador who lives
round the corner.
5.4. Constituent order in subclauses

The typical orderings of constituents described at the very beginning of this
chapter (S P, S P O, S P C, etc.) are the ones found in independent declara-
tive sentences. The orderings found in independent non-declarative sentences
- interrogative, exclamatory, imperative - were described in sections 4.4.4
Constituent order in subclauses 131
and 5.2.2. We must now take a look at constituent order in subclauses.

Before going into specific details, we can state as a very general rule that in
subclauses a w/i-constituent occupies the initial position irrespective of its
syntactic function. This can be illustrated by the following examples in
which the ννΛ-constituent does not realize a subject:
(1) The house which I've bought is beyond repair, (relative)
(2) I can't hear what you're saying, (interrogative)
(3) I remember what a mistake I made, (exclamatory)
5.4.1. Relative clauses

In relative clauses we find initial, relativized objects or complements:
(1) The honour (which) you do me is too great. ( d S P Oi)
(2) The man (that) you do this honour is my uncle. (Oi S P d)
(3) He is not the man (that) he was. (C S P)
As appears from the bracketing, the pronoun realizing an object or
complement may be omitted in restrictive relative clauses. In such cases Ο or
C is assumed to be realized by 'zero' (cf. section 4.2). In so-called non-
restrictive relative clauses, which do not restrict the reference of the
preceding noun or noun group but offer additional information, initial objects
and complements (as well as subjects and prepositional complements) are
always realized by overt pronouns. This can be illustrated by examples like
Susan, who(m) you met last night, has been promoted to sergeant and He
behaved like a child of twelve, which evidently he still felt himself to be.
If a relative pronoun functions as complement in a preposition group
realizing an adverbial, it is placed at the beginning of the clause, either
preceded by the preposition or followed by it at the very end of the clause:
(4) I called the editor to whom you posted the manuscript. (A S P O)
(5) I called the editor (who(m)/thai) you posted the manuscript to. (A- S P O -A)
As appears, a relative pronoun realizing a prepositional complement may be
omitted in restrictive clauses if the preposition is placed finally.
In those relative clauses which we term independent because they do not
function as dependent in a group but perform an independent clause function,
we also find objects, complements and adverbials in initial position:
(6) They believed what I told them. ( d S P Oi)
(7) That's exactly what it is. (C S P)
(8) Home is where you needn't mask your feelings. (A S P O)
5.4.2. Interrogative clauses

In subordinate wh-inierrogative clauses, we find the same constituent order
as in relative clauses. Objects, complements and adverbials realized by a wh-
word are placed initially as they are also in main-clause wA-interrogative
sentences, for example What are you doing these days?, What was the source
of her sorrow? and Where do you live? (see section 5.2.2 above). But in
subordinate w/t-clauses the order of subject and predicator is straight:
(1) I fail to see what you mean. (O S P)
(2) Tell me what your name is. (C S P)
(3) I don't understand how you can say that. (A S P O)
If an interrogative pronoun is complement in a preposition group, it is placed
at the beginning of the clause, preceded by the preposition or followed by it
at the end of the clause:
(4) I would like to know to whom I should post the manuscript. (A S P O)
(5) I would like to know who(m) I should post the manuscript to. (A- S P O -A)
In subordinate yes-no interrogative clauses, which begin with one of the
subordinating conjunctions if or whether, there is no fronting of objects,
complements or adverbials, nor is there inversion of subject and operator as
in main-clause yes-no questions (see section 5.2.2 above):
(6) He asks if we think the weather will stay fine. (SUB S P O)
(7) I wonder whether I hurt her feelings. (SUB S P O)
5.4.3. Exclamatory clauses

In subordinate exclamatory clauses, the order of constituents is the same as
in main-clause exclamatory sentences (see sections 4.4.4 and 5.2.2 above):
(1) It is impressive how confidently you work the machine. (A S P O)
(2) I remember what a silly mistake I made on that occasion. (O S P A)
As the order of subject and predicator is straight in subordinate wh-
interrogative clauses, subordinate exclamatory and interrogative clauses have
the same constituent order and therefore cannot be distinguished formally.
For example, a sentence like / told them how confidently Miss Tang worked
the machine is ambiguous. If it means Ί told them that Miss Tang worked the
machine with surprising confidence', the subclause is exclamatory. But if it
means Ί informed them about the degree of confidence with which Miss
Tang worked the machine', it is interrogative. In spoken language the
difference is signalled prosodically. If the subclause is intended by the
speaker to be exclamatory, the word confidently will be pronounced with
Constituent order in subclauses 133
strong stress and the rest of the clause with weak stress and low pitch. If it is
meant to be interrogative, on the other hand, not only confidently but also the
words Tang, worked and machine will be given prosodic prominence.
5.4.4. Other finite, nonfmite and verbless clauses

In other subclauses, constituent order in English (unlike that in many other
languages) is basically the same as in declarative sentences. Infinite sub-
clauses we find the same structures as the ones listed at the very outset of this
chapter, apart from the fact that there is often an initial subordinator:
(1) If you had phoned me I could have warned her. (SUB S P O)
(2) I guess she is Catholic. (S P C)
(3) It is unlikely that anything could happen to Maggie. (SUB SPA)
First, as noted in sections 5.3.3 and 5.3.4, there is sometimes partial
inversion in subclauses to signal conditional commitment (e.g. Had I known
this, I would never have accepted the offer) or as a result of attraction to a
clause-initial constituent (e.g. Alice is far more intelligent than are most of
her classmates).
Occasionally, in so-called concessive clauses with one of the conjunctions
as, if, that, or though, we find fronting of a complement (as in example (4))
or a full predication stack (as in example (5)). The fronted element(s)
appeals) to the left of the subordinating conjunction:
(4) Confused though he is, he manages beautifully, (cf. Though he is confused,
he manages beautifully.)
(5) I knew it for an aspect of my inheritance that I could never root out, deny it
though I might, (cf...., though I might deny it.)
In these examples, the fronted constituent receives more prominence than in
the corresponding non-inverted constructions.
In nonfinite subclauses we may also find the same constituents and
orderings as in declarative sentences:
(6) The case was closed by the defendant paying the fine. (S P O)
(7) Her oration finished, she breathed heavily with an overflow of indignation.
(SP)
(8) I want Gerard to become a schoolteacher. (S P C)
(9) She felt the earth shake violently. (S P A)
Typically, however, nonfinite clauses contain fewer constituents than finite
clauses. If the subject of the nonfinite verb form is understood to be identical
with the subject of the predicator in the matrix clause, it is not expressed:
(10) After shaking everyone's hand, the young lady nodded to several youths
standing behind huge reflectors. (P O)
(11) You must try to see things in perspective. (Ρ Ο A)
In so-called verbless clauses, i.e. clauses without a predicator, the only con-
stituents found are subordinators, subjects, complements and adverbials:
(12) With Peter on vacation, there is not much we can do. (S A)
(13) Although invariably polite, she was disliked by many. (SUB A C)
While the conjunction although in example (13) is a subordinator linking the
verbless clause to the matrix clause, -with in example (12) is interpreted as a
preposition which takes the verbless clause Peter on vacation as its com-
plement. It is therefore completely outside this clause and functions as head
of an adverbial preposition group.
5.5. Position and order of adverbials

In this chapter we have so far been concerned with the position and order of
obligatory sentence functions, especially in terms of deviation from the basic
sentence structures defined in sections 3.2.3-4. We shall now examine the
position and order of optional adverbials, i.e. the one major function type not
included in our list of sentence structures. As will be recalled (from section
3.2.9), optional adverbials are not restricted to one particular position in the
clause (which raises the question of what positions they may occupy), and
sometimes more than one adverbial may in fact be placed in the 'same'
position relative to other clause functions (which raises the question of
relative ordering of adverbials). Below we shall deal with these questions.
5.5.1. Main positions in finite clauses

In finite clauses we distinguish three main positions: initial (I), medial (M)
and terminal (T), as in the following examples:
(1) Quite frankly, she gave him every excuse.
(2) Owen substantially improved my abilities as a student.
(3) The Voice had not been idle for the summer.
Each of these main positions needs closer examination.
5.5.2. Initial position

I-position is always before the subject (or the predicator plus subject in
inverted constructions) but after conjunctions, if any:
(1) Quite frankly, did she give him any excuse?
Position and order ofadverbials 135
(2) I admire her because, quite frankly, she gave him every excuse.
(3) He loved her and, quite frankly, she gave him every excuse.
Initial adverbials may occur after a fronted constituent (e.g. an object):
(4) This solution, obviously, she would never accept.
In relative subclauses, initial adverbials follow the relative constituent if it
realizes a clause function other than subject (e.g. object):
(5) He drew a lesson from it which, in his heart, he had long understood.
5.5.3. Medial position

Any adverbial which follows the subject but precedes the head verb of the
predicator is said to be in M-position. This definition accommodates several
more specific M-positions:
Pre-M operator central-M aux2 post-M1 aux3 post-M2 ... H:v
The following examples illustrate these different M-positions:
(1) Owen substantially did improve my abilities as a student.
(2) Owen had substantially improved my abilities as a student.
(3) Owen may have substantially improved my abilities as a student.
(4) Owen's abilities may have been substantially improved, too.
Note that the adverbials are not in fact restricted to one specific M-position:
in each example the adverbial could be moved to at least one other M-
position without this affecting the acceptability of the construction (e.g.
Owen may substantially have improved my abilities as a student as a possible
alternative to (3)).
In example (1), the adverbial is in pre-M position after the subject and
before the first auxiliary, the operator, on whose presence in the clause a pre-
M adverbial is dependent. This position is often marked in BrE and requires
primary stress on the operator. In AmE, pre-M position is often unmarked
(see below).
In example (2), the adverbial is in central-M position: it is the unmarked
M-position after the subject and the operator, if any, and before any other
auxiliaries. Thus in both Owen substantially improved my abilities ... and
Owen had substantially improved my abilities the adverbial is said to be in
central-M position. In the first example, the distinction between pre-M and
central-M is neutralized but since the position of the adverbial is as
unmarked as in the second example, we classify it as central-M. The most
frequent type of adverbial, the clause negator NOT, always appears in
central-M position:
(5) Owen did not improve my abilities as a student.

(6) Owen could not improve my abilities as a student.
In negative interrogative constructions, NOT is contracted across the subject
to pre-subject position in spoken and informal written English; compare:
(7) Had/j 't Owen improved my abilities as a student?
(8) Had Owen not improved my abilities as a student?
Moving on to examples (3) and (4), we find that they illustrate two different
post-M positions: post-M1 after the second auxiliary in a string and post-M2
after the third auxiliary in a string. Post-M2 is relevant only after the past
participle of BE in passive and progressive constructions (as in example (4)
and e.g. Owen may have been actively trying to undermine her authority).
What all the four different M-positions seem to share is immediate 'contact1
with the predicator somewhere before the head verb. There is an exception to
this, involving the primary verb BE:
(9) She is always happy.
(10) Keith was clearly a man of the world.
The adverbials in these examples are classified as central-M adverbials. BE
being an operator in such examples (cf. section 3.2.1), we expect central-M
adverbials to follow it rather than precede it. This means that the following
constructions, which are marked in BrE (unlike AmE) and require primary
stress on the primary verb, illustrate pre-M position:
(11) Jane never" was fond of her sister.
Note also that, in negative constructions, not follows rather than precedes BE
(which supports the characterization of the position as central-M):
(12) She is not happy.
Although it is possible to find similar examples with HAVE followed by not
(e.g. We haven't the time), as a primary verb it usually behaves like an
ordinary füll verb with respect to the position of adverbials:
(13) Keith never had what it took to be a murderer.
(14) Kathy obviously had other ideas.
Pre-M position is common in elliptical clauses with an operator representing
a full predicate:
(15) I miss you, darling, I really do.
(16) You know that I fancy you, I always have.
Post-M2 is statistically rare. This position is relevant in passive and
progressive constructions, where, typically, manner and degree adjuncts in
M-position immediately precede the head verb, to which it is closely related

semantically, irrespective of the number of auxiliaries:
(17) She was seriously crippled by the blow. (central-M)
(18) She may be secretly supporting their cause. (post-M1)
(19) He may not have been completely recovering from the accident. (post-M2)
The precise M-position assigned to such adjuncts thus depends on the
number of auxiliaries.
5.5.4. Terminal position

Turning now to T-position, we observe first of all that it is the position of
obligatory adverbials in S P A and S P O A constructions:
(1) Last week I stayed with my parents.
(2) She reluctantly stood the figure on a stool.
T-position is also the position of optional adverbials which follow objects,
complements and obligatory adverbials, or intransitive predicators:
(3) She gave me a quick kiss on the cheek.
(4) Her parents got very upset when they saw us together.
(5) I put the gun in my pocket as casually as I could.
(6) He smoked incessantly.
In transitive and complex-transitive constructions we sometimes encounter
cases where the optional adverbial follows the predicator (and hence is not in
M-position) but immediately precedes an object or obligatory adverbial:
(7) Many subscribers to NLLT experiencedybur times a year excitement of a
sort that the arrival of a scholarly journal in one's mailbox rarely occasions.
(8) I put the gun casually in my pocket.
The position taken up by the italicized adverbials in these examples is called
pre-T. Pre-T is the position after Ρ but before other obligatory constituents.
In clauses with both a provisional and a real subject, the real subject
normally follows adverbials in T-position if the provisional subject is
realized by it:
(9) It worried me immensely that Jack did not get in touch.
If the provisional subject is realized by there, the real subject is more tightly
integrated in the clause structure (cf. section 3.2.2) and usually precedes
adverbials in T-position:
(10) There will be two loaded guns in the top drawer.
As with Ο A sequences (see above), considerations of weight (and semantic

clarity) may force an adverbial into pre-T position:
(11) There once lay, in the valley, a large boulder, reputedly the first chunk of
America to be touched by the Pilgrims' feet.
Similarly, adverbials which most naturally occur in central-M position are
sometimes placed in pre-T position, as in the following cases involving
copula verbs and subject complements:
(12) His parents seemed often very dependent on his support.
(13) Jack became only very rarely this angry with his wife and children.
Pre-T position is thus primarily to be thought of as a more marked alternative
position where adverbials with a potential for straight T-position or some M-
position are sometimes placed (see further sections 5.5.8 and 5.5.11 below).
5.5.5. Overview of positions in finite clauses

Let us briefly summarize the seven different positions in which we find
adverbials:
I Quite frankly, she gave him every excuse.
pre-M Owen substantially Mid improve my abilities as a student.
central-M Owen substantially improved my abilities as a student.
Owen had substantially improved my abilities as a student.
post-M1 Owen may have substantially improved my abilities as a student.
post-M2 Owen's abilities may have been substantially improved, too.
pre-T I put the gun casually in my pocket.
Τ She gave me a quick kiss on the cheek.
5.5.6. Positions in nonfinite clauses

In nonfinite clauses: a) adverbials do not occur in genuine I-position; b) the
distinction between pre-M and central-M is neutralized because there is no
operator (operators being always finite, cf. section 3.2.1); c) pre-T and Τ are
parallel to pre-T and T in finite clauses (e.g. Having put the gun casually in
my pocket, I opened the window and She reassured me by giving me a quick
kiss on the cheeK\.
Although superficially adverbials do occur initially in nonfinite clauses,
they never actually co-occur with a subject:
(1) Always feeling guilty, she mostly kept to herself.
(2) Obviously having difficulty reading the book, he decided to watch television
instead.
The italicized adverbials here correspond to central-M adverbials in finite

clauses containing a subject:
(3) She was always feeling guilty.
(4) He was obviously having difficulty reading the book.
When a subject is present in a nonfinite clause, it precedes such adverbials:
(5) Sarah always feeling guilty, I did not want to tell her what had happened.
(6) We all felt depressed, Jack obviously having difficulty reading the book.
Moreover, for clausal negation, NOT immediately precedes the first verb in
the predicator, even BE:
(7) Not feeling guilty, she never hesitated to call on them.
(8) Not being happy with this decision, he quit.
It is therefore more appropriate to classify the adverbials in all these
examples as M-adverbials, more specifically as central-M adverbials, than to
classify them as I-adverbials.
The second characteristic feature of adverbial positions in nonfinite clauses
is the absence of a distinction between pre-M and central-M adverbials. By
definition, pre-M adverbials are adverbials that precede the operator. Since
there is no operator in nonfinite clauses, there can be no pre-M adverbials,
either. However, on analogy with finite predicators, adverbials are often
placed after the first auxiliary in nonfinite predicators. We thus find
unmarked adverbials both before and after the first auxiliary of a nonfinite
predicator — even in BrE; compare the following examples:
(9a) With Jack always being so honest, we endanger the whole operation.
(9b) With Jack being always so honest, we endanger the whole operation.
We classify always in the a-example as a central-M adverbial and always in
the b-example as a post-M1 adverbial, but the difference between the two
positions is not as clearly felt as in finite clauses.
There is in nonfinite predicators normally a maximum of two auxiliaries,
the second of which is always the past participle of BE. Like post-M2 in finite
clauses, post-M2 in nonfinite clauses is relevant only in passive or
progressive constructions:
(10) Having been repeatedly counteracted by his superiors, he decided to quit.
(11) To have been deliberately misrepresenting things in this way, she must have
been mentally deranged.
To complete the picture, there is a special M-position in nonfinite predicators
between infinitive marker and infinitive. An adverbial in this position creates
the so-called split infinitive (for discussion, see e.g. Quirk et al. 1985: 496ff):
(12) For me to suddenly resign my job is unthinkable.

(13) She has tried to consciously stop worrying.
The split infinitive is felt by many to be not only marked but in fact un-
acceptable. Despite widespread prejudice against it, this construction type is
by no means rare. In some cases, as in (13) above, it even serves the purpose
of making the meaning more precise and unambiguous; compare (13) to the
following examples with continuous ίο-infinitive constructions (cf. Quirk et
al. 1985: 497):
(13') She has tried consciously to stop worrying.
(13") She has tried to stop consciously worrying.
5.5.7. Factors governing the distribution of adverbials

There are few hard-and-fast rules for the distribution of adverbials in the
positions defined in sections 5.5.1-6 above. The fact that many adverbials are
mobile often makes it impossible to predict the exact location of a particular
adverbial in a particular clause. Thus, except for the negation NOT and a few
other adverbs (typically adverbs of degree and frequency) such as ALWAYS,
EVER, OFTEN, NEVER, RARELY, REALLY, HARDLY (which are nearly
always located in central-M position in BrE), it is more appropriate to speak
of certain factors contributing to the speaker/writer's location of adverbials.
Though T-position is for many adverbials (especially obligatory adverbials
and long optional adverbials) the most natural, unmarked position, there is a
great deal of variation. In the following sections we shall look briefly at
some of the most important factors contributing to this variation. They are: 1.
the form and relative weight of the adverbial; 2. (con)textual cohesion; 3. the
scope of the adverbial; 4. semantic clarity; 5. stylistic considerations.
5.5.8. Position, form and relative weight

The form of the adverbial is the major factor determining the position it
assumes in the clause. Statistical analyses have shown that long adverbials
(clauses and long groups) tend to occur in I- and especially T-position (in
accordance with the principle of end-weight) rather than in M-position:
(1) Knowing that Keith would be elsewhere, I staked out the Black Cross.
(2) One of the black guys was staring at me with either affection or contempt.
Short adverbials (adverbs and small groups) are more evenly distributed in
the three main positions:
(3) I cried all the time, actually.
(4) Actually I cried all the time.
(5) I actually cried all the time.

Within the range of different M-positions, even short adverbials are
relatively rare in pre-M and especially post-M.
M-position is especially rare for weighty adverbials if the subject is an
unstressed personal pronoun. Thus while we find:
(6) Guy, feeling no closer to life than to death, pressed on.
we do not find:
(7) *I, knowing that Keith would be elsewhere, staked out the Black Cross.
Long parenthetical adverbials occasionally occur in M-position:
(8) Keith, i/i my mind, blew all his chances.
(9) Nicola Six, he confided to all who cared to listen, sang in the church choir.
In constructions with complex predicators such adverbials assume pre-M
position, even in BrE:
(6*) Guy, feeling no closer to life than to death, was pressing on.
(81) Keith, i/i my mind, had blown all his chances.
As already indicated in section 5.5.4, considerations of weight may lead to
adverbials in pre-T position, preceding direct objects:
(10) She drank two cups of black coffee and tasted with hunger the black tobacco
of a French cigarette.
(11) How do I know, for instance, that Keith works as a cheat?
Such ordering is especially frequent when the object contains a dependent
clause. Even if the adverbial is actually weightier than the object, we usually
get A O rather than Ο A in such cases:
(12) She thought, absolutely wrongly and with characteristic lack of imagination,
that they loved her for her mind.
In the passive counterpart to a ditransitive construction, the adverbial by-
phrase will typically appear in pre-T position if the direct object is a clause or
lengthy group:
(13) Keith was told by various magistrates that he had a 'poor character1.
5.5.9. Position and (con)textual cohesion

As noted above, T-position is for many adverbials the most unmarked
position, i.e. the position they take up unless there is some reason for taking
up a different position. (Con)textual cohesion (i.e. the binding together of
sentences in a coherent text or, in speech, the use of utterances in particular
contexts) is one such reason for appearing in I-position rather than T-

position. As with other clause functions, adverbials appearing initially either
receive special highlighting (as a fronted constituent) or establish a link to
the context or preceding text, thus forming part of what is 'given' (cf. section
5.2.3). In sections 5.3.4 and 5.3.6 we dealt with cases involving fronting of
adverbials accompanied by partial and full inversion, respectively. However,
adverbials may appear in I-position to receive highlighting without this
resulting in inversion. Thus instead of simply saying e.g.:
(la) You are going to have a smashing time in Copenhagen.
we may put more focus on the adverbial in Copenhagen by saying:
(Ib) In Copenhagen you are going to have a smashing time.
The subtle difference between the two examples can be formulated in this
way: in the first example, the information contained in the adverbial in
Copenhagen forms part of the prediction expressed by the clause as a whole,
whereas in the second example, this information forms part of a given
background to the prediction even if it does not link up explicitly with the
preceding context. It is this background effect of initial adverbials that
explains why many adverbials in I-position do in fact link up very nicely
with the preceding context, as in the following examples:
(2) Sometimes, when he stumbled into her bedsit in the small hours, Analiese
was alone. On other occasions he surprised her in bed with famous people.
(3) When Analiese gave herself to you, she would give herself utterly, and
probably wouldn't ring the house. In this last particular alone, appearances
were deceptive.
In many such examples the link to the preceding context is created
specifically by anaphoric pronouns (such as the demonstratives this and that
and referential it) or other definite expressions.
Not surprisingly, I-position is thus the 'natural' (but not the only) position
for adverbials serving more particularly as conjuncts (cf. section 3.2.9):
(4) Besides, Keith generally preferred short girls.
(5) To conclude, I'll never have dealings with Intercom again.
5.5.10. Position and scope

By the 'scope' of an adverbial is meant the extent of its semantic relations to
other constituents or to the context. The scope of an adverbial may be just
one other constituent, typically the predicator, as in:
(1) He walked briskly down the road.
or it may be a whole sentence:

(2) To tell the truth, he was limping when he got up off the basketball court.
There is in language a general tendency for the scope of constituents to be
determined by their position: the earlier they occur in the linear sequence the
broader their scope. There is thus a narrowing of scope as we move from left
to right in a clause or sentence, according to what is sometimes referred to as
the principle of linear modification (Solinger 1952, for discussion see
Johansson & Lysväg 1991, vol. II: 240ff). The varying scope ofadverbials is
very nicely demonstrated by comparing examples which differ with respect
to the position of the adverbial:
(3a) Clearly he saw her.
(3b) He saw her clearly.
(4a) Quite frankly he told me about the affair.
(4b) He told me quite frankly about the affair.
Here the adverbials in the a-examples are disjuncts with sentential scope ('It
was clear that he saw her' and am telling you quite frankly that he told me
about the affair'), while the adverbials in the b-examples are likely to be
interpreted as manner adjuncts with predicator scope. As in these examples,
adverbials in I-position tend to have wider scope than adverbials in M- or T-
position. Thus disjuncts (such as initial To tell the truth, Clearly and Quite
frankly in the examples above) and conjuncts (see examples in section 5.5.9)
are frequently in I-position (or M-position), while adjuncts relating more
narrowly to the predicator or the predication are frequently in M- or T-
position. The difference of scope is, however, also seen in adverbials of the
same kind (see Johansson and Lysväg 1987, vol. II: 241), as in the following
example where a place adjunct is placed in I- and T-position, respectively:
(5) Outside in the street we could see the car. (i.e. 'we were in the street1)
(6) We could see the car outside in the street, (i.e. 'the car was in the street')
When disjuncts with sentence scope are placed in T-position, they are felt to
be more detached from the rest of the clause, like an afterthought:
(7) I wasn't worrying, for Christ's sake.
(8) She fell for the old trick, obviously.
In such cases the adverbial is always separated by the rest of the clause by
an intonation boundary, if emphatic and stressed.
A few conjuncts appear in T-position, especially in short sentences:
(9) I didn't want to go anyway.
(10) She wouldn't sleep with him, though.
5.5.11. Position and semantic clarity

In any communication, a very basic consideration on the part of the speaker
or writer is to make sure that constituents which are intended to belong
together semantically are placed in such a way that they are recognized by
the listener or reader as actually belonging together. Thus, for example, if we
wish to communicate the information that someone decided to study art and
that this decision was made in Paris (to use an example from Johansson &
Lysväg 1991, vol. II: 241), (la) is more appropriate than (Ib):
(la) In Paris he decided to study art.
(Ib) He decided to study art in Paris.
Example (Ib) is likely to be interpreted to mean that the studying of art was
to take place in Paris: in Paris seems to locate the 'studying' rather than the
'decision-taking'. But strictly speaking, (Ib) could mean the same as (la) and
is thus ambiguous. Syntactically, the difference between the two readings of
the (Ib) hinges on whether or not in Paris belongs as an adverbial to the
object clause to study art. In (la), the adverbial is unambiguously a matrix
clause function. To ensure an unambiguous reading, the speaker or writer
may thus decide to place the adverbial in I-position.
In transitive constructions where we normally expect the order A, the
form and weight of the object may lead to a reversal of this order (see sec-
tions 5.5.4 and 5.5.8). But semantic clarity is also an important consideration
in such constructions (cf. Quirk etal. 1985: 499, Dienhart 1992: 122f):
(2) She herself interviewed with hurtful disdain the student I had turned down.
(3) He urged secretly that she be dismissed.
(4) He hoped fervently to be applauded.
In these examples, the only alternative to pre-T position, if we want to
preserve the intended meaning, is M- or I-position, not T-position, which
would drastically change the meaning of the sentence.
The distribution of adverbials within the range of M-positions is sometimes
governed by considerations of semantic clarity (cf. Quirk et al. 1985: 494):
(5a) She hadn't really delighted her audience.
(5b) She really hadn't delighted her audience.
(6a) They have seriously considered him for the post.
(6b) They seriously have considered him for the post.
The italicized adverbials here function as disjuncts in pre-M position and as
degree or manner adjuncts in central-M.
5.5.12. Position and style

While other clause functions are relatively fixed in the basic sentence
structures (cf. sections 3.2.3-4 and 5.1), optional adverbials often move
around more freely and thus lend themselves to stylistic variation. For
example, in a string of S P O constructions, the monotony of the initial S in
this basic sentence structure can be broken by placing adverbials in different
positions relative to S. Thus, instead of letting the adverbials take up
unmarked M- or T-position, as in:
(1 a) She slowed her pace with obvious deliberation. She briskly removed her hat
and the black clip that secured her chignon. She shook out her hair with a roll
of her throat.
we could let some of them occupy I-position to create a little variation:
(Ib) She slowed her pace with obvious deliberation. Briskly she removed her hat
and the black clip that secured her chignon. With a roll of her throat she
shook out her hair.
The effect of this operation is to tie the sentences more closely together in a
more balanced and easy-going flow of narration.
Another stylistic factor is avoidance of heavy adverbial clusters. There is a
natural tendency for adverbial clustering in T-position in accordance with the
general principle of end-weight. However, the possibility of placing
adverbials in other positions makes it possible to avoid too heavy clusters in
one position and to create more evenly balanced constructions:
(2) On the fifth day the sun burst through again inexorably.
(3) In excellent fettle, in the pink or the blue of boyish good health during their
absence, Marmaduke sickened dramatically within a few hours of their
return.
Avoidance of clustering in a particular M-position may lead to utilization of
different M-positions; compare:
(4a) The key-note address was certainly elegantly delivered by David.
(4b) The key-note address certainly was elegantly delivered by David.
In negative clauses, where NOT occupies central-M position, disjuncts may
be shifted into pre-M position:
(5) She probably has not seen David yet.
(6) Keith obviously will not listen to you.
Marked, or unexpected, adverbial position may be a device characteristically
employed by authors to create a certain effect, mood or atmosphere in a
novel, as arguably in the following constructions containing adverbials in
pre-T position (for discussion, see Dienhart 1992: 123ff):
(7) He crumpled some bread in a bowl and poured over it hot milk.
(8) On Sunday he got up early and took, from the suitcase his serge suit.
5.5.13. Relative position of adverbials

When adverbials appear in clusters, as they frequently do at I-position (for
extensive backgrounding) and especially at T-position (in accordance with
the general principle of end-weight), the question of sequential order be-
comes relevant. Again there are no hard-and-fast rules but rather a number of
general tendencies. These tendencies can be described in terms of some of
the factors governing the position of adverbials reviewed above: form of real-
ization, relative weight, scope, semantic and stylistic considerations. Often
the order is determined by the interaction of several of these factors.
A) Order and form/relative weight. There is a tendency for short
adverbials to precede long adverbials. Single words thus often precede
groups, and groups often precede clauses:
(1) They were walking [arm in arm] [towards the cafe where they had taken to
having their mid-morning snack].
(2) [In excellent fettle], [in the pink or the blue of boyish good health during
their absence], Marmaduke sickened [dramatically] [within a few hours of
their return].
(3) [Twenty minutes later], [as he strode back up the beach], the wind threw
everything it had at him.
Identity of form is often avoided in clusters of adverbials (cf. Quirk et al.
1985: 649):
(4) ΊFinally probably only John will be chosen.
(5) ?I kissed her on the cheek on the plane.
B) Scope. From the point of view of scope and the principle of linear
modification, we would expect the sequence disjuncts-conjuncts-adjuncts
(e.g. adjuncts of time, place and manner) in I-position. A sequence of all
three types is possible but cumbersome (cf. Quirk et al. 1985: 651):
(6) [Unfortunately], [however], [last night] the old lady declined to see me.
Disjuncts tend to precede adjuncts in I-position but inversion is possible -
especially if the disjunct serves as a parenthetical insertion; compare:
(7a) [To be quite honest], [last night] he vowed never to see her again.
(7b) [Last night], [to be quite honest], he vowed never to see her again.
There is more vacillation in sequences containing both a conjunct and an
adjunct:
(8a) [However], [last night], he vowed never to see her again.

(8b) [Last night], [however], he vowed never to see her again.
In T-position adjuncts generally precede conjuncts and disjuncts (which are
often felt to be added as parenthetical afterthoughts):
(9) Keith left the pub [a bit later], [unfortunately].
(10) Trish called her [twice], [probably].
Sequences of all three types of adverbial (adjunct-conjunct-disjunct) are
possible in T-position but rare:
(11) She didn't like him [much] [anyway], [to be frank].
In a sequence of adjuncts in T-position, those with narrow predicator scope
precede those with broader predication or sentence scope:
(12) Hope kissed him [passionately] [on the cheek].
(13) Keith will react [violently] [if we press him too hard].
C) Semantic considerations. Closely related to the question of scope is the
question of semantic content. Adverbials in clusters often supplement each
other progressively so that each new adverbial in the sequence offers further
elaboration, determination or specification:
(14) [On the front passenger seat], [under the elegant rag of a white silk scarf],
lies a heavy car-tool.
(15) [Right from the start], [from the moment that her thoughts began to be
consecutive], Nicola knew two strange things.
Sometimes, adverbials seem to be in a completely random order, simply
enumerating different aspects:
(16) A couple of mornings a month, [stiff with pride], [deafened with aspirin],
[reckless with Bloody Marys], Nicola would adumbrate serious reform.
(17) [Feeling neither vigour nor its opposite], [feeling no closer to life than to
death], [feeling thirty-five], Guy pressed on.
Time and place adjuncts are occasionally followed by adverbial clauses
either relating some event to the time/place expressed by the adjunct or
bringing the plot or narration forward:
(18) I'd been standing under the sign saying TAXIS [for about a half-hour] [when
the royal-blue Cavalier made its second circuit and pulled up at the bay].
(19) Keith pondered and agonized [for several days] [before filling in the section
marked HOBBIES].
There is a general tendency for time and place adjuncts to precede adjuncts
expressing circumstance, purpose or contingency:
(20) He had entered the competition [some months ago] [on the advice of various
friends and admirers].
(21) The boulder had to be moved [closer to the shore], [in order to satisfy
expectations of how history ought to happen].
Many grammarians have noted that we often find the sequence manner-
place-time-others (cf. e.g. Steller & S0rensen 1974: 105; Preisler 1992: 60):
(22) She played [beautifully] [at the concert] [last night] [even if she hadn't had
time to practice].
There is great vacillation, however:
(23) He whispered the same words [at night] [in the hotel] [with that strange
accent of his].
(24) He called her [at three o'clock in the morning] [from the Black Cross].
In M-position time and frequency adjuncts usually precede manner adjuncts:
(25) He had [never] [knowingly] drunk a glass of wine.
(26) She had [often] [secretly] admired Keith.
In T-position after predicators expressing motion, short manner adjuncts tend
to precede adjuncts expressing place or direction:
(27) Keith rose [steadily] [towards the very crest of his new profession].
(28) She walked [evenly] [to the back].
Similarly short manner adjuncts tend to precede obligatory adverbials:
(29) I put the gun [casually] [in my pocket].
Note also the position of the short manner adjuncts in the following cases:
(30) I think [with shame] of my contorted little crib in Hell's Kitchen.
(31) I was looking [with dread] at the blood on your hands.
(32) Keith dealt [angrily] with her application.
The italicized constructions are either independent adverbials or part of a
transitive construction involving a discontinuous prepositional verb (THINK
of, LOOK at, DEAL with) followed by a direct object (see section 4.3.2).
5.6. Discontinuity
It has been pointed out earlier that the forms realizing a constituent tend to
stay together rather than to be interrupted by other constituents but that this
principle is often defeated by other principles (illocutionary value, high-
lighting, increasing complexity). In the following sections we shall examine
the types of discontinuity found in English and the conditions under which
discontinuity occurs.
Discontinuity 149
5.6.1. Discontinuous clauses

Arguably we have discontinuous clauses in cases like:
(1) John, I think, never returned to Ireland.
(2) This, he claimed, was caused by the inefficient handling of young criminals.
These sentences contain parenthetical 'matrix clauses' with verbs of convic-
tion or communication. It is tempting to analyse such sentences in terms of
discontinuous object clauses, as in e.g.:
STA
John never returned to Ireland
This proposal reflects our interpretation of the discontinuous clause as

somehow the 'logical' object of the verb of conviction or communication. By
adopting this analysis we capture the obvious relationship between such
examples and continuous alternative expressions like:
(!') I think (that) John never returned to Ireland.
(21) He claimed (that) this was caused by the inefficient handling of young
criminals.
In these examples, it seems, the only possible analysis of the italicized part is
O:cl. Note, however, that there is a significant difference between the two
sets of examples: unlike the continuous object clauses, the discontinuous
ones cannot be supplied with an explicit marker of subordination (e.g. * That
John, I think, never returned to Ireland). Thus while the status of the clause
as a subclause is clear in the continuous examples, it is at least questionable
in the discontinuous examples. Another, related problem with the proposed
analysis of the discontinuous cases is that it does not reflect the parenthetical
status of the 'matrix'.
If we want to reflect the parenthetical nature of the 'matrix', this can be
done by assigning to it instead the status of dependent, adverbial clause in
relation to the rest of the sentence, which accordingly is analysed as a dis-
continuous clausal function stack:
John never returned to Ireland
This analysis is supported by the fact that the parenthetical clause, like many
adverbials (e.g. probably), can be deleted without this affecting the status or
grammaticality of the rest of the sentence (John never returned to Ireland).
The only way to avoid recognizing discontinuity in such examples is to
treat the parenthetical clauses simply as fully integrated adverbials (like e.g.
never). But such an analysis is clearly less revealing of the relationships
involved than the other proposals.
In relative subclauses there is further evidence for an interpretation
involving discontinuous clauses in connection with verbs of conviction or
communication. Consider the following examples:
(3) The house that you told me Jack built last year is now put up for sale.
(4) The passage which he thought he had referred me to had been removed
The discontinuity is here caused by the relativization and fronting of an
object or prepositional complement, as in the last example, where we have
two discontinuities (a clause and a preposition group):
A- he thought S
he had referred me
which to
Discontinuity 151
Unless clearly marked as parenthetical (by means of intonation or e.g.

commas: The passage which, he thought, he had referred me to ...), the
subject and predicator are here not as readily interpreted as parenthetical as
in main clauses and therefore should not be assigned adverbial status.
In interrogative sentences, we get discontinuous clauses in cases like:
(5) When did you say he would arrive?
(6) Why did he say she left her husband?
(7) How did you say she packed the suitcase?
Such sentences are strictly ambiguous: the question relates either to the first
predicator or to the second, i.e. 'when did you say1 or 'when would he arrive';
'why did he say1 or 'why did she leave her husband'; 'how did you say' (e.g.
angrily or impatiently) or 'how did she pack the suitcase'. If the question
relates to the first predicator, there is no clausal discontinuity:
QUE
d
P- s -P Ο
adv g- pro -g cl
Why DEP he H she left her husband

V V
did say
If, however, the question relates to the second predicator (i.e. if the
underlying question is 'why did she leave her husband?', rather than 'why did
he say anything about it?'), we have a discontinuous clause:
QUE
cl
P- S -P -Ο
cl- g- pro -I
Γ
-cl
^\·
^ \
DEP he ]Ά S 1
adv V v Fιτο Λ
Why did say she left her husband
The interesting feature of such interrogative sentences is that the question

word (e.g. why) triggers (or attracts) subject-operator inversion in the matrix
clause, not the subclause, no matter which of them it relates to semantically.
Thus, even when the interpretation is such that we assume an underlying

question 'why did she leave her husband' rather than 'why did he say
anything about it?', we do not actually say * Why he said did she leave her
husband. This, however, fits nicely with the lack of subject-operator
inversion in corresponding continuous interrogative subclauses:
(8) Did he say why she left her husband!
Having discussed discontinuous finite clauses, we turn now to discontinuous
nonfinite clauses. Corresponding to cases with verbs of conviction or
communication like John, I think, never returned to Ireland, we get nonfinite
clauses with a fronted, topicalized constituent of the following type:
(9) Linda he believed to be less innocent than Jim.
This sentence is derived from He believed Linda to be less innocent than
Jim. Notice that the subject-predicator sequence he believed cannot here be
interpreted as parenthetical: *Linda, he believed, to be less innocent than
Jim. There is thus no possibility of analysing he believed as an adverbial.
Topicalization of a constituent in a nonfinite object clause may cause
discontinuity, irrespective of the nature of the verb in the finite predicator:
(10) This vase I want you to sell to the Americans.
(11) The old lady they saw climbing the stairs with difficulty.
Such examples (derived from / want you to sell this vase to the Americans
and They saw the old lady climbing the stairs with difficulty) are fairly
marked. Embedded in relative clauses, however, where such topicalization is
inbuilt, as it were, we find unmarked cases:
(12) You won't like the essay which you wanted me to submit.
()
Ο you wanted S _P
pro pro g
which me to submit
Discontinuity 153
As pointed out in section 5.2.3, part of a message may be postponed in order

to highlight it and achieve end-weight. This was illustrated by an example
like It worries me that the children have not returned (Sp P O Sr), in which
the clause realizing the subject is replaced by the substitute form it and
moved to the end of the sentence. Extraposition, however, is not the only
way in which end-focus and end-weight can be achieved. Instead of moving
an entire subject clause to the end of the sentence it is possible to divide it
and place the part of it following its subject in final position. This happens
with to-infinitive clauses:
(13) He is believed to be guilty.
(14) She is known to behave rather badly.
(15) / happen to know him well.
(16) John seems to take his job too much to heart.
(17) Alfred appears to be hungry.
(18) His explanation was hard to believe.
Such discontinuity (traditionally referred to as 'nominative with infinitive') is
found in passive sentences (such as (13) and (14)) with verbs like ALLOW,
BELIEVE, EXPECT, FIND, HOLD (in the sense of 'consider'), KNOW,
SUPPOSE and THINK, in active sentences (such as (15) to (17)) with verbs
like APPEAR, HAPPEN, SEEM and in sentences with complements (such as
(18)) realized by adjectives (or adjective groups) like DIFFICULT, HARD,
EASY, SIMPLE. While extraposition of an entire subject clause is largely
optional (compare examples like It would be awkward to go now and To go
now would be awkward), final placement of the predicator and any following
constituents of ίο-infinitive clauses is obligatory in examples like the ones
given above. For instance, structures like *He to be guilty is believed and
*Alfred to be hungry appears are ungrammatical.
While end-focus and end-weight are achieved both by means of
extraposition and by means of discontinuity, the latter strategy serves the
additional purpose of highlighting one of the functions of the subclause, e.g.
the subject, through placement initially in the sentence. For example, the
speaker gives more prominence to Alfred in choosing Alfred appears to be
hungry than It appears that Alfred is hungry.
It is not entirely self-evident that it is the entire ίο-infinitive clause which
functions as subject in the type of examples discussed here, and that
discontinuity is thus in fact involved. It would not be far-fetched to assume
instead that it is the (pro)nominal element preceding the finite verb which
functions as subject, i.e. that the subject of an example like She is known to
behave rather badly is realized by the pronoun She exclusively. In favour of
such an analysis it could be argued that the subject of a nonfinite clause
normally takes the objective case (as in / want him to go) and that this is not
the case here. Furthermore, there is concord of number and person, not
between the ίο-infinitive clause and the finite verb but between the initial
element and the finite verb. The type of construction we find is / happen to
know him well, not *I/Me happens to know him well, and as subject clauses
count as 3rd person singular for concord, this is an argument against a
clausal subject analysis. Thirdly, it is clearly the (pro)noun alone, not the
combination of the (pro)noun and infinitive construction which constitutes
the topic of the utterance.
Semantically, however, it is clearly the state of affairs described by the
entire ίο-infinitive clause which is specified by the finite verb. In He is
believed to be guilty, for example, it is not 'he1 but 'he to be guilty' that is
believed. In favour of a clausal subject analysis it may be argued syntact-
ically that there are closely related structures in which it is rather obviously
the entire clause which functions as a sentence constituent. Corresponding to
the passive sentences there are often active sentences in which a continuous
TO-infinitive clause functions as the object ('accusative with infinitive'):
(13') The judge believes him to be guilty.
(14') We know her to behave rather badly.
Corresponding to the active sentences, similarly, there are sentences with
extraposition in which a continuous thai-clause functions as the real subject:
(15') It (so) happens that I know him well.
(16') It seems that John takes his job too much to heart.
(17') It appears that Alfred is hungry.
For these reasons a clausal subject analysis is here preferred to a non-clausal
one. That the subject of the discontinuous ίο-infinitive clause takes the
subjective rather than the objective case and determines the number and
person of the finite verb is assumed to be due to the special condition that it
appears in the position normally reserved for the subject of finite verbs. The
choice of the subjective rather than the objective case seems to be motivated
by position rather than function.
Since it looks as if the subject of the subclause has been 'raised' into the
matrix clause - note that it displays the defining subject characteristics A
through D mentioned in section 3.2.2 - the term subject raising is used by
many grammarians in their account of sentences of the type discussed in this
section.
Discontinuity is occasionally also found in passive sentences in which the
predicator of the subject clause is an -ing participle, compare:
(19) The girl was heard screaming.
Discontinuity 155
(19') They heard the girl screaming.

As the goal against which the event described by heard is directed appears to
be the event described by the girl screaming rather than the person referred
to by the girl, it is reasonable to assume that the object is here clausal. In this
analysis, the subject of the corresponding passive sentence may be assumed
to be the discontinuous -ing participle clause The girl... screaming.
5.6.2. Discontinuous verb groups

Verb groups are very frequently discontinuous. In the first place this is the
case in all interrogative sentences except yes-no questions with BE as main
verb, wA-questions in which the interrogative pronoun realizes (part of) the
subject and tag questions (in which the main verb is ellipted):
(1) Did she ever spend the money I gave her?
(2) Isn't there supposed to be a reading here in about twenty minutes?
(3) How would you suggest that I get in touch with her?
Discontinuous verb groups are found also in negative sentences, in which the
operator is separated from the rest of the verb group by the negative adverb
NOT, and in sentences where other adverbials are placed after the operator:
(4) You haven't missed much.
(5) Other than the title, I don't know a thing about your book.
(6) I've never trusted hand squashers.
(7) I'm only joking.
A much less important source of discontinuity is the fronting of predication
stacks (cf. section 5.3.7):
(8) [Standing around in the shops] were heavily bearded men with long
gaberdine coats, wearing thick boots.
Finally, verb groups can be viewed as discontinuous in cases like the
following, which involve complex predicators (cf. section 4.3):
(9) Mary called the man up.
(10) Miranda waited diligently on the Wilson family.
(11) They put her behaviour down to lack of confidence.
5.6.3. Discontinuous noun and pronoun groups

While premodifiers are hardly ever separated from their heads in noun
groups, postmodifiers are sometimes postponed in order to observe the
principles of end-focus and end-weight, i.e. for reasons of information
structure. Since postmodifying clauses are typically longer and heavier than
postmodifying groups, it is mainly clausal postmodifiers which are separated

from their head noun or pronoun. As illustrated by the following examples,
these may be so-called elaborative f/raf-clauses (cf. section 10.1.3), relative
clauses, -ing participle clauses and to-infinitive clauses:
(1) The rumour spread that the King had been beheaded.
(2) Peter met a girl/someone last night who lives in Tasmania.
(3) I saw a woman yesterday carrying a dead baby in her arms.
(4) The time has come to evict the squatters from our premises.
In (1) and (4) the subject noun (or pronoun) group is so long and heavy and
the rest of the sentence (the part realizing the predicator) so short and light
that the principle of increasing complexity would be grossly violated if the
postmodifying clauses were not postponed, i.e. these sentences would be
completely off balance without discontinuity. In (2) and (3), on the other
hand, discontinuity could be avoided without throwing the sentence off
balance by moving the adverbial to the initial position, e.g. Last night Peter
met a girl/someone who lives in Tasmania. But if the speaker does not wish
to give prominence to this part of the message, and if Peter constitutes his
communicative point of departure (the topic, see section 3.2.2), such fronting
would run counter to his communicative intentions.
Like postmodifying clauses, postmodifying groups may be separated from
their head noun/pronoun in order to achieve end-focus or for reasons of end-
weight. As appears from the next examples, both preposition groups and
nominal modifiers may be postponed in this way:
(5) The entire crew/Everybody was drowned except the captain of the ship.
(6) What business is that of his!
(7) I pressed the trigger and a hole appeared in his forehead the size of a quarter.
Postponed preposition groups often denote exception and begin with except,
but, save, excluding or apart from.
Not only clauses and groups but also single words may be separated from
the noun/pronoun they postmodify. For example, emphatic self-forms
(himself, herself, etc.) can readily be moved from the position right after the
noun to the end of the sentence to achieve end-focus:
(8) Helen/She told me the sad news herself.
(9) The boys/They have often made that mistake themselves.
The degree of emphasis provided by the self-form decreases with distance
from the (pro)noun it postmodifies, compare:
(1 Oa) John/He himself read the book several times.
(1 Ob) John/He read the book himself several times.
Discontinuity 157
(1 Oc) John/He read the book several times himself.

The first of these examples, where there is no discontinuity, is the most
emphatic, and the third example, where the distance between head and
dependent is the greatest, is the least emphatic (unless the pronoun is
pronounced with very heavy stress).
Like self-forms, the quantifying indefinite pronouns all, both and each can
be separated from their head noun to give prominence to them:
(11) The students have all understood what you are driving at.
(12) My brothers have both understood what you are driving at.
(13) My brothers have each taken a big apple.
In these examples, all and both are separated from their head through
postponement to the position after the operator of the verb group (compare
All the students / Both my brothers). In the example with each, the only
corresponding continuous construction is a pronoun group (Each of my
brothers/*Each my brothers) or a singular noun group (Each brother).
Interestingly enough, the separate position typically occupied by these
quantifiers is the adverbial central-M position (i.e. between S and P if P is
realized by a full verb, or after the operator if P is complex or realized by the
primary verb BE; cf. section 5.5.3):
(14) The students (probably) all understood what you are driving at.
(15) The students may all have understood what you are driving at.
(16) The students were all interested in what you are driving at.
Note that each, unlike all and both, allows adverbial T-position:
(17) My brothers took a big apple each.
The possibility of inserting probably between The students and all in the first
example shows that the pronoun does not simply behave like a post-
modifying dependent in a continuous subject group.
On the evidence provided by examples (14) to (17), a possible alternative
to an analysis in terms of discontinuity is to interpret all, both and each when
separated from their heads as adverbials.
In pronoun groups we find the same positions for all, both and each as in
noun groups:
(18) They (probably) all understood what you are driving at.
(19) We have both understood what you are driving at.
(20) They have each taken a big apple.
(21) They have taken a big apple each.
Notice that continuous constructions (as in All the boys, Both my brothers,
etc.) are not possible with pronominal heads. The only option for the speaker
is to promote the dependent pronoun to head in a pronoun group; compare:
(22a) *All they understand what you are driving at.
(22b) All of them understand what you are driving at.
(23a) *Boih we understand what you are driving at.
(23b) Both of us understand what you are driving at.
In all the constructions with all, both and each discussed here, the function of
these pronouns is to provide a further specification of the quantification of
the head noun/pronoun. An alternative way of offering such specification,
especially in spoken or informal English, is right-dislocation, i.e. a kind of
delayed apposition, where a parenthetical dependent construction is used
which is capable of replacing the head:
(24) The boys left the party, both of them.
(cf. Both of them left the party)
(25) They understood what you are driving at, all of them.
(cf. All of them understood what you are driving at)
In this way, both and all can receive end-focus like each (cf. The boys took a
big apple each). Unlike most other cases of appositional dependents,
continuity is only possible if the head of the construction is a pronoun rather
than a noun, or if the apposition is clearly marked as parenthetical:
(26a) They both of them left the party.
(26b) *The boys both of them left the party.
(26c) The boys - both of them - left the party.
(27a) They all of them understood what you are driving at.
(27b) *The students all of them understood what you are driving at.
(27c) The students - all of them - understood what you are driving at.
Again it is possible to view the quantifying expression as an adverbial rather
than as a dependent in a discontinuous group.
Leaving quantifying dependent pronouns aside, we turn finally to cases of
internal discontinuity in noun groups:
(28) The discovery in 1929 of penicillin has saved millions of lives.
While the subject noun group is not interrupted by another constituent here,
the normal order the discovery of penicillin in 1929 has been changed to give
end-focus - in the first of the two tone groups which this utterance is divided
into at normal speech tempo - to the word penicillin. In noun groups of this
type, the two dependents are not coordinated. What we find here is
subordination, for while of penicillin is a dependent of discovery only, in
Discontinuity 159
7929 is a dependent of the discovery of penicillin. This is shown in the

following tree diagram:
H_ DEP
χ
PEP DEP in 1929

art g
The discovery of penicillin
This order reflects the normal S P O A order in the clause underlying the
nominalization: Someone discovered penicillin in 1929. If the dependent
preposition group in the head stack the discovery of penicillin is moved to the
position after in 1929, this head noun group is realized discontinuously. This
is shown in the next tree diagram. Communicatively, the effect of such
internal discontinuity is the same as that of many instances of external
discontinuity. In either case a part of a message is given extra prominence
through postponement.
The discovery of penicillin
5.6.4. Discontinuous preposition groups

In wA-questions and relative clauses, the wA-complement of a preposition
often leaves the preposition behind at the end of the clause when it is fronted
to clause-initial position:
(1) What did you do that/o/?
(2) My computer, which I've had a lot of trouble with, is up for sale.
In independent declarative sentences, preposition groups may be realized

discontinuously if the speaker wishes to topicalize the referent of the
complement:
(3) This well you draw water from.
For examples like Democracy must be fought for every day containing a
stranded preposition, see section 4.3.2.
Finally, it should be mentioned that there is double discontinuity in an
example like the following:
(4) Kiri is wonderful to work with.
This sentence may be regarded as derived from It is wonderful to work with
Kiri, which in its turn - through extraposition of the subordinate TO-
infinitive clause - may be considered derived from To work with Kiri is
wonderful. The communicative purpose it serves is to topicalize Kiri, and
this is done by moving the complement of the preposition group realizing the
adverbial in the subclause into sentence-initial position (where it ousts the
substitute form //). As a result of this fronting both the subject clause and the
adverbial preposition group are realized discontinuously, the former as Kiri
... to work with and the latter as Kiri ... with.
STA
Kiri with
5.6.5. Discontinuous adjective and adverb groups

Adjective and adverb groups are often realized discontinuously in compar-
ative constructions. We shall therefore begin this section by examining some
of the basic facts of adjective and adverb comparison in English. We operate
with three members of the category of comparison: positive, comparative and
superlative. There are two types of formal expression: morphological (or
Discontinuity 161
inflectional, or synthetic) comparison and syntactic (or analytic) comparison.

The following table summarizes the possibilities:
Positive Comparative Superlative
Morphological tall taller tallest

comparison
early earlier earliest
Syntactic beautiful more beautiful most beautiful

comparison
wisely more wisely most wisely
A maximal comparative construction consists of three parts: a comparative

element, an adjective or adverb, and a comparative basis. The comparative
element is either realized morphologically as an inflection of the adjective or
adverb (-er, -est) or syntactically as a separate word (the adverbs more/less
or most/least). The comparative basis is the standard on the basis of which
the comparison is expressed. This is typically realized as a ίΑαη-construction
in the comparative degree and an (»/-construction in the comparative or
superlative degree:
(1) Sally is taller than I am/than me.
(2) Mary danced more beautifully than Jane (did).
(3) John is the brighter of the two.
(4) Sam worked the most diligently of them all.
The sentence analysis of morphologically compared adjectives or adverbs is
straightforward:
STA
Sally
taller than I am
STA
John
the brighter of the two
While the o/-construction is always considered a group, the than-

construction may sometimes be interpreted as a clause with than as
SUBiconj (e.g. than I am) or as a group with than as Hrprep (e.g. than me).
The sentence analysis of syntactically compared adjectives or adverbs is
more complicated and may well involve discontinuity. In constructions like
Sally is more beautiful than Jane and Jane danced more beautifully than
Sally, we might simply treat the comparative element (more) and the
comparative basis (the ίΑαη-construction) as separate dependents of the head
adjective or adverb. But there are two alternatives. The first alternative is to
group the comparative element and the adjective or adverb together as a H:g
followed by the than -constructions as a DEP:g or DEPicl. This solution is
parallel to constructions with morphologically compared adjectives and
adverbs (i.e. where the comparative element is morphologically 'grouped
with1 the adjective or adverb). The second alternative is to treat the
comparative element and the constructions expressing the comparative basis
as a discontinuous dependent of the head adjective or adverb; e.g.:
STA
Sally
more than Jane
A similar analysis could be postulated for superlative constructions like Sam

worked the most diligently of them all with the most ... of them all as a
discontinuous dependent of the head adverb diligently.
Discontinuity 163
An argument in favour of treating the construction expressing the com-

parative basis as a dependent of the comparative element rather than of the
head adjective or adverb is that it provides a specification of the comparative
element. Furthermore, its occurrence is more directly dependent on the
occurrence of the comparative element. Thus we cannot say e.g. *Sally is
beautiful than Jane or *Sam worked diligently of them all. And if we leave
out the construction expressing the comparative basis, its meaning is implied
only if the comparative element is present. Thus Sally is more beautiful
implies e.g. than Jane whereas Sally is beautiful does not. It therefore seems
reasonable to assume that there is a close relationship between the com-
parative element and the construction expressing the comparative basis, and
an obvious way to reflect this in our analysis is to treat them as grouped
together in a discontinuous constituent.
Another kind of discontinuity occurs in comparative constructions when
another clause function - usually an adverbial - intervenes between the
adjective or adverb and the construction expressing the comparative basis:
(5) Sally is calmer now than she was yesterday. (S P C- A -C)
(6) Mary danced more beautifully last night than ever before. (S P A- A -A)
In the first example the adjective group realizing the subject complement is
split into two parts by an adverbial, and in the second the adverb group
realizing the adverbial (of manner) is split by another adverbial (of time).
These discontinuities are motivated by considerations of end-weight and end-
focus, i.e. they are to do with information structure.
If a comparative adjective group containing a comparative basis modifies a
noun, it is often realized discontinuously:
(7) Sally is a calmer woman than she was last year.
If sufficiently weighty, such constructions may be realized as a continuous
postmodifier in relation to the noun:
(8) Sally is a woman noticeably far more considerate than Mary.
We also find discontinuous adverb groups modifying adjectives:
(9) John is better prepared than his sister.
Not only comparative forms but also superlative forms may be interpreted as
discontinuous in examples like the following:
(10) Peter has the strongest will of them all.
(11) This is the least dramatic anecdote of the ones I've heard.
In such cases, however, the status of the definite article is problematic: does
it go with the superlative form or with the noun?
When the comparative basis is an OF-construction we sometimes find topic-

alization of this construction with obvious discontinuity as a result:
(12) Of all the boys Peter was by far the brightest.
(13) Of the two teams Norwich were the more efficient.
We also find discontinuity in constructions expressing comparison which are
neither comparative nor superlative:
(14) Sally is as calm now as she was yesterday. (C:g)
(15) Mary danced so beautifully last night that people began to cry. (A:g)
(16) The person who called me was too busy at the time to answer my questions
over the phone. (C:g)
(17) SO ardently had he been preaching the efficacy of prayer that he now silently
invoked the name of Diana for every desire that passed through his head.
(A:g)
In examples (14) to (16), the construction expressing comparison is realized
discontinuously with an intervening adverbial. In (17) the head and
premodifier of the italicized adverb group have been fronted in order to
highlight that part of the message, and as the adverbial begins with so this
fronting is accompanied by partial inversion (see section 5.3.4). In all four
examples, the discontinuity is clearly due to factors of weight and balance.
Discontinuous adjective groups are not only found in comparative
constructions. Adjective groups containing heads like comparable (to ...),
different (from ...), difficult (to ...), easy (to ...), opposite (to ...), (impossible
(for, to ...), preliminary (to ...), similar (to ...) ana suitable (for...) are often
realized discontinuously:
(18) It falls into a different category from the rest.
(19) I don't imagine she was an easy person to live with.
(20) This was only a preliminary little gathering to the one planned for the
autumn.
Considerations of weight and end-focus may lead to continuous post-
modification instead:
(21) It falls into a category very different from the rest.
(22) I don't imagine she was a person in any way easy to live with.
(23) This was only a little gathering supposedly preliminary to the one planned
for the autumn.
6. Coordination and subordination
In this chapter, we shall examine two major syntactic relations more closely
than we have done so far: coordination and subordination. We begin by
recapitulating some of the basic facts about these two relations (see also
sections 3.3.3 to 3.3.5).
6.1. Introduction
In English, as in other languages, linguistic units can be linked in two
different ways: subordination and coordination, as shown in the following
examples, in which the units linked together are clauses:
(1) I'll tell him that you called.
(2) His hair was rumpled and he wore a raincoat over his pyjamas.
In (1) the two clauses have a different syntactic status: the second is em-
bedded in the first, as a realization of its direct object. Subordination is here
signalled by the subordinating conjunction that. In (2) the two clauses have
the same syntactic status. Neither is subordinated to the other and embedding
is not involved. Coordination is here signalled by the coordinating conjunc-
tion and.
Coordination and subordination apply not only to clauses but also to
groups. The following examples illustrate subordination in groups:
(3) I have formidable enemies in the Literature Division.
(4) It's all quite friendly.
(5) He helped an old lady across the street.
In the italicized groups in (3) and (4) the dependents formidable and quite are
subordinate to the heads enemies and friendly, respectively. We also operate
with subordination in preposition groups. Thus in (5) the preposition across
is analysed as head and the prepositional complement the street as sub-
ordinate dependent (cf. section 3.3.2).
Coordination of words and groups can be illustrated by:
(6) Henry and Charles were talking only yesterday about how brusque and
boorish he's become, (noun + noun, adjective + adjective)
(7) I never thought I'd be surprised at anything he did or said, but he and I had a
very odd conversation the other day. (verb + verb, pronoun + pronoun)
(8) Did your album sell over or under 10,000 copies? (preposition + preposition)
(9) On his side there were/tower arrangements and corny get-well cards, (noun
groups)
166 Coordination and subordination
(10) This is very unfortunate and acutely embarrassing, (adjective groups)

(11) What he has said or may have done in the past need not concern us at this
stage, (verb groups)
As is apparent from these examples, it is frequently words of the same class
which are coordinated. In an example like Jim and I, we have coordination of
words from closely related word classes (nouns and pronouns).
So far we have looked at examples of binary coordination or binary sub-
ordination, i.e. examples with a relation between two linguistic units only.
But in both coordination and subordination the number of linguistic units
linked together need not be restricted to two. The following examples
illustrate multiple coordination:
(12) On our return we were cold and hungry and exhausted.
(13) The woman had earlier drunk what one witness called a "lethal cocktail" of
cider, vodka and Drambuie.
(14) Would you like port or Madeira or claret?
(15) I waited and worried and hoped and prayed.
In sentences with the coordinating conjunctions and and or there is in
principle no limit to the number of conjoints that may be joined together. In
such sentences - where the conjunction is commonly 'understood1 except
between the last two conjoints - coordination is thus recursive: we can go on
adding conjoints. Nothing in the structure of English prevents this, but in
actual communication the speaker will generally avoid taxing the patience
and memory of the listener with immoderate multiple coordination.
Multiple subordination is illustrated by the next examples:
(16) In very great poetry the music often comes through even when one doesn't
know the language.
(17) Anything overtly religious filled him with pagan alarm.
In (16) the head of the italicized noun group is poetry, and the dependent is
the adjective group very great in which great functions as head and the
adverb very as dependent. In (17), we have a pronoun group with Anything as
head and overtly religious as a dependent group consisting of religious as
head adjective and overtly as dependent adverb.
As in the case of coordination with and and or, there is in principle no end
to the amount of subordination which can be used in English sentences, i.e.
subordination is recursive. Here as well, however, massive multiple linking
will tend to be avoided in order to prevent monotony and difficulty of
understanding due to memory limitations (cf. e.g. the cat on the mat in the
kitchen in the house by the river that...), see also section 3.3.5.
Coordination 167
6.2. Coordination
6.2.1. Coordination and ellipsis
In coordination it is sometimes the case that one or more words are left out
from one of the conjoints. Such ellipsis can be illustrated by the following
example (see also section 4.2):
(1) Tanner is our dean and Stopford our bursar.
As linguistic units joined together in coordination are normally formally
similar, we assume that the verb is has been omitted between Stopford and
our bursar and that it is therefore clauses which are coordinated:
STA
Tanner our dean Stopford our bursar
Once ellipsis has been recognized in some cases of coordination, we must

decide under what conditions such a zero-analysis is warranted. Consider:
(2) Henry is very brusque and boorish.
Here there is no need to postulate ellipsis as in (1), for the conjoined units are
syntactically alike, both with respect to form and function:
STA
Henry is
very CJT CO CJT

adj conj adj
brusque and boorish

It would not be unnatural to assume, however, that the adverb very has been
omitted between and and boorish. If this analysis is chosen, it is adjective
groups rather than adjectives which are coordinated. But the example can be
expanded even more:
(2') Henry is very brusque and (Henry is very) boorish.
If such an analysis with three zeros is adopted, the conjoints are neither
adjectives nor adjective groups but clauses. The question which has to be
considered is now how much ellipsis we should recognize in coordination?
If we adopt a strategy of maximal expansion, we end up in a situation where
coordination involves only clauses. An example like:
(3) Did your album sell over or under 10,000 copies?
will then be analysed as Did your album sell over (10,000 copies) or (did
your album sell) under 10,000 copies? with zero realization of all the words
in parentheses. Such an analysis would seem to be more complicated than
necessary, and in a sentence like He was a good, honest, kind, gentle bobby
doing what the Met does best, doing what British policing does best, working
with his local community, it becomes inordinately complex. To this we
should add that sometimes expansion into clauses is not even possible. While
a sentence like John and Mary lived in Cambridge last year can be expanded
in such a way that the conjoints become clauses, John and Mary met in
Cambridge last year cannot, for when used intransitively MEET is a
reciprocal verb requiring reference to at least two persons. So while we can
say John lived in Cambridge we cannot say John met in Cambridge. For the
same reason we cannot operate with ellipsis and clausal coordination in a
sentence like John and Mary are very much alike with a reciprocal adjective.
Note also that e.g. The woman had earlier drunk what one witness called a
"lethal cocktail" of cider, vodka and Drambuie cannot be dealt with in terms
of clausal coordination due to the fact that a cocktail has to be mixed.
Finally, maximal expansion gives problems with concord of number between
subject and predicator. If there is assumed to be clause coordination in a
sentence like John and Mary live in Cambridge, we have to select a different
form of the verb in the expanded structure with ellipsis: John (lives, in
Cambridge) and Mary lives, in Cambridge.
For reasons like these we shall operate with ellipsis only when the
"missing form" is retrievable without this affecting the analysis of the other
constituents and when the specification of a function without an overt form
clarifies the overall structure. In the example John and Mary live in
Cambridge we thus do not operate with ellipsis, partly because this would
affect the form of the verb and partly because it would have no clarifying
Coordination 169
effect. Nor is there assumed to be ellipsis in example (2) above (Henry is

very brusque and boorish). On the other hand we do recognize ellipsis in an
example like (1) (Tanner is our dean and Stopford our bursar). Here the
conditions just mentioned are met. If we did not operate with a missing form
(is), the conjoints of the coordinate structure would be syntactically
dissimilar. Note also that the second conjoint cannot stand for the whole.
While we can say Tanner is our dean, we cannot say *Stopford our bursar.
6.2.2. What can be coordinated?

In section 6.1 we showed that clauses may be coordinated with clauses,
groups with groups and words with words. Often, however, we find
examples of what looks like mixed coordination:
(1) No doubt they think it aristocratic and very grand.
(2) What he said or may have done on that occasion need not concern us now.
What we find here is coordination of a word and a group, i.e. of formally
different units. Nevertheless, the conjoints are formally similar in the sense
that the head word of the group belongs to the same class as the single word:
in (1) we have adjectivals and in (2) we have verbals (cf. section 3.3.1 on the
use of these form terms). Furthermore, the conjoints are functionally equi-
valent, not only in the obvious sense of being both conjoints but, more
significantly, by being each capable of assuming the same function as the
compound unit as a whole. In (1) either of the conjoints may thus by itself
realize the object complement and in (2) either of the conjoints may by itself
realize the predicator in the subject clause. Even in cases where the conjoints
of a compound unit cannot single-handedly assume the function of the unit
as a whole (as in e.g. John and Mary met in Cambridge), the conjoints are
felt to be functionally of exactly the same status.
So far, then, the conjoints of a coordinated structure are functionally
identical and formally similar. But they may also be formally dissimilar:
(3) She played the sonata expertly and with great spirit.
(4) This will not explain the fundamental magic of his personality or why I still
have an overwhelming wish to see him.
In (3) the conjoints are realized by an adverb and a preposition group and in
(4) by a noun group and a clause. In order to avoid coordination of
unmatching entities, some grammarians might prefer to operate with ellipsis
and clausal coordination here. Though formally different, however, the
conjoints are functionally identical: in (3) either of the conjoints is capable of
realizing the adverbial by itself, and in (4) either of the conjoints is capable
of realizing the object by itself. For this reason we shall therefore operate
here with formally mixed coordination rather than ellipsis. If we assumed

instead that there was clausal coordination, the 'missing forms' would be
retrievable without this affecting the analysis of the other constituents, but
the resulting expressions would be very cumbersome and there would be
little clarification of structure.
We also find coordination of stretches of speech which are neither words
nor groups nor clauses:
(5) My sister lives in Norwich and works for the council.
(6) He might drop into a bar and down some liquor.
In example (5) we have coordination of predicates and in example (6) we
have coordination of predications. In section 4.1, we introduced the notion of
stacking to cope with such examples. Would it be more reasonable to analyse
them in terms of clause coordination? Such an analysis would yield:
(5') My sister lives in Norwich and (my sister) works for the council.
(6*) He might drop into a bar and (he might) down some liquor.
Intuitively this solution is not very convincing, and in some cases it is not
even feasible. In an example like the following with both-and it is simply not
possible to insert the 'missing' words, for this correlative conjunction (to be
discussed in section 6.2.4 below) cannot link finite clauses:
(7) This solution both satisfies our staff and placates our students.
Here we cannot insert this solution after and:
(71) "This solution both satisfies our staff and (this solution) placates our
students.
For this reason we assume that in some cases the conjoints of a coordinated
structure are stacks of constituents, i.e. collections of constituents somehow
belonging together without constituting any of the basic grammatical form or
function types recognized in this book.
Our approach to coordination is thus to recognize ellipsis but to operate
with it very sparingly. In those cases where the conjoints are formally non-
equivalent but functionally identical in the sense discussed here, as in
examples of the type illustrated by (1) through (4), ellipsis is not recognized.
Nor is there assumed to be ellipsis in examples like (5) to (7), where the units
joined together are not words, groups or clauses but stacks. On the other
hand we do recognize ellipsis in examples of the type Tanner is our dean
and Stopford our bursar and John likes apples and Bob cherries. As pointed
out in section 6.2.1, the second conjoint could not stand for the whole here,
and there is therefore no viable alternative to an analysis by which the
predicator has been omitted and both the conjoints are clauses. In short, only
Coordination 171
in those cases where one conjoint differs from the other in not being able to
stand for the whole is coordination assumed to involve ellipsis.
Finally we should point out that the conjoints of a coordinated structure are
nearly always semantically similar. For clauses to be coordinated they must
have a common theme, as in His hair -was rumpled and he wore a raincoat
over his pyjamas (note that an example like His hair was rumpled and he
lived in Cambridge challenges us to think of a context where 'rumpled hair1
and 'living in Cambridge' are thematically related). In example (1) (No doubt
they think it aristocratic and very grand), the semantic function of either
conjoint is to describe an attribute - or characterization - of the referent of
the object pronoun it. In an example like (3) (She played the sonata expertly
and with great spirit), where a word and a group are coordinated, the
semantic function of either conjoint is to describe the manner in which the
sonata was played. Occasionally, however, we come across coordination of
semantically incongruous units. This can be illustrated by the following ex-
ample (see Matthews (1981: 213), by whose work this chapter is influenced):
(8) He left in a Rolls Royce and a bad temper.
Here the semantic roles of the conjoints are markedly different. Together
with the preposition the former describes the means by which the person
referred to left and the latter the state in which he left. Such coordinations are
exceptional, though, and often characteristic of humorous style. This being
so, we can round off this section by stating that the units which can be
coordinated always have the same function, nearly always express the same
type of meaning and frequently have the same form.
6.2.3. Types of coordination

As pointed out in section 3.3.3, there are two types of coordination
depending on whether the coordinator is present or understood: linked and
unlinked. The following example illustrates three instances of linked
coordination:
(1) I never thought I'd be surprised at anything he did or said, but he and I had a
very odd conversation the other day.
Linked coordination is clearly the normal state of affairs. In those cases
where the conjoints of a compound unit are adjectives, adjective groups,
adverbs or adverb groups, however, coordination is not infrequently
unlinked. The next examples illustrate unlinked coordination of adjectives:
(2) He was speaking in a low, urgent voice.
(3) Reason is always apparent to the discerning eye. But luck? It is invisible,
erratic, angelic.
In unlinked coordination the missing coordinator practically always corres-

ponds to and. In tree diagrams and linear analyses such an understood and
could be represented by the symbol 0, but this would be redundant.
In linked multiple coordination there is typically only one overt co-
ordinator (see section 3.3.3). Its position is between the last two conjoints:
(4) They found Guzman vain, garrulous and excessively concerned about his
health.
(5) Would you like port, claret or Madeira?
While and and or may thus be omitted between the first conjoints in a series,
and while and may be omitted altogether as in examples (2) and (3), the
coordinator but is never omitted from a coordinated construction.
Besides being linked or unlinked, coordination can be divided into two
types depending on whether the order of the conjoints is reversible or
irreversible. In coordination, unlike subordination, the units joined together
have the same syntactic status, and it is therefore not surprising that they can
sometimes be interchanged with little or no difference of meaning:
(6a) Stopford is deaf to Tanner's indignation and Wilson tends to ignore Cook's
accusations.
(6b) Wilson tends to ignore Cook's accusations and Stopford is deaf to Tanner's
indignation.
Conjoints realized by words or groups are also frequently interchangeable.
For example, the ordering of the prepositions is immaterial in an example
like Did your album sell over or under 10,000 copies? In many cases,
however, the order of the units joined together is fixed. In the first place,
coordinated clauses cannot be interchanged if the second describes an event
which is subsequent to or follows from that described by the first:
(7) She slammed the door and Philip pulled out from the kerb with a short cry of
rubber, ('and then')
(8) I have forgotten my keys and there is nothing I can do to help you. ('and
therefore')
Nor is reversal of order possible if the second clause of a compound unit
contains a pro-form that refers back to a word in the first clause, or if there is
ellipsis in the second clause:
(9) The new constitution will enable Fujimori to stand for election in 1995, and
it will give him the power to close congress again.
(10) John likes apples and Bob cherries.
Like clauses, coordinated verbs or verb groups often describe successive
events, cf. an example like He came, saw and conquered. Here the order of
the conjoints is fixed as well. In those cases where other words or groups are
Coordination 173
joined together, the ordering of the units is in principle free. But reversal of
the order in which such conjoints occur may be obstructed by the principle of
end-weight mentioned in section 5.2.3. In an example like This is sad and
acutely embarrassing the conjoints of the compound unit realizing the
subject complement cannot be interchanged without throwing the sentence
off balance. Furthermore, such a reversal would invite a reinterpretation of
acutely as modifying both embarrassing and sad.
Coordination may be recursive or non-recursive. As pointed out in section
6.1, there is in principle no limit to the number of conjoints that may be
joined together by and and or. Coordination with but, however, is non-
recursive. Here the number of conjoints is always restricted to two, and
structures of the type */ like claret but not port but Madeira do not occur. In
an example like // wasn't cheap, it wasn't easy, but it's the best solution, the
understood coordinator between the first two clauses is and, not but:
STA
CJT but it was the best solution

0
It wasn't cheap it wasn't easy
A distinction must also be drawn between simple and complex coordination.

Simple coordination involves functionally unitary constituents: by itself each
conjoint would serve only one clause or group function. Thus, in e.g. My
neighbour and his daughter laughed, each of the conjoints might serve as the
subject (My neighbour laughed and His daughter laughed). By contrast, in
cases of complex coordination, each conjoint contains a form stack of
constituents (see section 4.1.2) serving more than one function:
(11) He gave Barbara a kiss and Ann some good advice.
(12) Laurence called my brother a fool and my sister a slut.
(13) She sold and / bought the house.
The conjoints in these examples are form stacks, each containing more than
one function: indirect object + direct object (example (11)), object + object
complement (example (12)), and subject + predicator (example (13)). By
definition coordination of predicate stacks (as in My sister lives in Norwich
and works for the council} and of predication stacks (as in He might drop
into a bar and down some liquor) is also complex coordination.
Additional information is often added to a clause in the form of an
appended construction consisting of a coordinator and a conjoint:
(14) Barbara sings beautifully, and Joan too.
(15) She is brilliant, but not her husband.
(16) He has threatened to leave her, or sue her parents.
Such appended coordination typically involves elliptical clauses.
Note finally that we sometimes have coordination of completely identical
constituents. In such constructions there is no notional coordination:
(17) Jack became more and more upset.
(18) They talked and talked and talked.
(19) Well, you know, there are teachers and teachers.
This is called pseudo-coordination. In the first two examples the effect of
pseudo-coordination is intensification. In (19) the expression implies that
there are different kinds of teacher (e.g. good teachers and bad teachers).
6.2.4. Coordinating conjunctions

In all the examples of coordination given in this chapter so far, the
coordinators have been realized by and, or or but (or there has been assumed
to be an and or or missing). This is no coincidence, for these words are the
only ones in English which are unproblematically classifiable as coordinating
conjunctions. An important property by which they differ from other linkers
is that they are restricted to the position between the units they connect. In
this way they differ from subordinating conjunctions, which besides
occupying the position initially in the second clause can also introduce the
first of two linked clauses:
(la) The children want a holiday abroad, whereas we would rather stay at home.
(Ib) Whereas we would rather stay at home, the children want a holiday abroad.
(2a) The children want a holiday abroad, but we would rather stay at home.
(2b) *But we would rather stay at home, the children want a holiday abroad.
In those cases where nouns or noun groups are connected, the three
coordinating conjunctions differ from prepositions in the same way: and, or
and but are fixed in the position between the conjoints, and immediately
preceded and followed by them, while prepositions such as with, in, on and
by can be placed in several positions together with their complements:
(3a) Patton and his army sped northwards.
Coordination 175
(3b) *And his army Patton sped northwards

(3c) (*) Patton sped northwards and his army
(4a) Patton with his army sped northwards.
(4b) With his army Patton sped northwards.
(4c) Patton sped northwards with his army.
Example (3c) is only acceptable if and his army is added as an appended
afterthought and even then the example becomes much better if we add too:
Patton sped northwards, and his army too.
Another property of and, or and but is that they cannot be preceded by a
conjunction the way subordinators sometimes can, for example whether in a
sentence like / don't know where she is or whether she has received my
telegram. The three words also differ from other linkers in permitting
ellipsis, for example in Tanner is our dean and Stopford 0 our bursar,
Tanner can be transferred or Stopford 0 fired and Tanner is highly qualified
but Stopford 0 a less controversial choice. There are only two ways in which
the three coordinators differ from each other: unlike and and or, but cannot
link more than two conjoints together and it cannot be omitted.
Some linking words are so similar to and, or and but that it would not seem
unnatural to include them among the coordinating conjunctions. Among the
clausal linkers the conjunction for comes particularly close to being a pure
coordinator. It can only stand between the units it connects and it does not
permit a preceding conjunction:
(5) Injection is the most dangerous way of taking drugs, for the drug goes
straight into the bloodstream.
Note, however, that ellipsis is not possible in clauses introduced by for the
way it is in examples like the ones with and. or and but given above. For
instance, the auxiliaries cannot be omitted in the second clause of Stopford
has been hired, for Tanner has been transferred in spite of the fact that they
are identical with the ones occurring in the first clause. Furthermore, the
conjunction for differs from and, or and but in being a purely clausal linker.
While for thus differs from the three prototypical coordinators, it is not
clearly classifiable as a subordinator either. What this shows us is that the
division of conjunctions into coordinators and subordinators is not clear-cut.
There is a grey zone with conjunctions which in some ways behave like
coordinators and in others like subordinators.
Another linker which resembles coordinating conjunctions is as well as,
which appears to perform the same function as and in examples like He is
clever as well as determined and Henry as well as Charles objected. When
as well as links nouns and noun groups together in the subject position,
however, the finite verb of a sentence in the present tense normally ends in -s
if the first noun (group) is singular: Henry as -well as Charles objects. In this
way as well as differs from and: Henry and Charles object. In triggering
singular rather than plural concord under these circumstances as well as
behaves unlike a coordinator. In Henry as well as Charles objects, the word
sequence as well as appears to subordinate Charles in relation to Henry.
Another group of words which resemble coordinating conjunctions are
adverbs such as YET, SO, STILL and NOR:
(6) He felt sorry for her yet at the same time relieved.
(7) I was furious, so I didn't mince my words.
While these words are clearly linkers and behave like coordinators in being
restricted to the position between the units they connect, they differ from
and, or and but in that they can be preceded by a conjunction. We therefore
assume that YET, SO, STILL and NOR are linkers which at the same time
serve as adverbials (more specifically as conjuncts) within the clause they
introduce.
Semantically, and is inclusive and non-contrastive, but inclusive and
contrastive and or exclusive and disjunctive, as illustrated by respectively
clever and reasonable, clever but unreasonable and clever or stupid. Besides
having additive meaning, and sometimes appears to signify 'subsequent to' or
'following from1, as in e.g. He knocked three times and went in and She felt
nervous and took a tranquillizer. These meanings - which can be made
explicit by inserting then or therefore after and - are not expressed by and,
however, but are due to the conjoints, which describe temporally or causally
related situations (and which are therefore irreversible). Similarly, or
sometimes seems to signal a negative condition (if not... then), for example
in You must put on winter woollies or you'll catch cold. This meaning -
which can be made explicit by inserting else after or - does not belong to or
either, though, but is due to the conjoints, which describe conditionally
related situations (and which are irreversible here as well). In order for but to
be used, finally, the conjoints must be semantically contrastive, and (as
pointed out in section 6.2.2) the conjoints of any coordinated structure must
have a common theme. Both these conditions are met in clever but unreason-
able, where the common theme is 'mental characteristics'.
There are three pairs of correlative conjunctions: both ... and, either ... or
and neither... nor:
(8) He decided both to leave his wife and to sell his Porsche.
(9) Baroness Thatcher is one of those politicians you either love or hate.
(10) Women priests tend to be neither authoritarian nor submissive.
We suggest that correlative conjunctions are analysed in this way:
Coordination 177
C_
cu
CO CJT CO CJT
conj a<y conj acjj
neither authoritarian nor submissive
Both ... and and either ...or can be regarded as expansions of and and or, and
like these they express inclusive and exclusive meaning respectively. They
differ semantically from and and or, however, in that they give extra
prominence to the individual conjoints, i.e. both and either underscore the
bipartite nature of the following unit. Distributionally, both ... and differs
from and in being unable to link finite clauses. For example we cannot say
*Both his hair was rumpled and he wore a raincoat over his pyjamas
(compare example (2) in section 6.1 above). Otherwise there are no restric-
tions on the type of conjoints it can connect.
While and and or can link more than two conjoints, multiple coordination
is not normally found if the conjunctions used are correlative. For example,
sentences like On our return I was both cold and hungry and exhausted and
Would you like either port or Madeira or claret? are unacceptable to most
speakers. In the case of either ...or and neither... nor multiple coordination
cannot entirely be ruled out, though. This is apparent from the following
authentic example:
(11) Why did God bestow on Homo sapiens such an abundance of emotions?
What is their biological function? Neither Plato nor Spinoza nor
Schopenhauer could really answer this.
Unlike and and or, but is not matched by any pair of correlative conjunc-
tions. However, it often combines with the adverb group not only:
(12) Not only was he gifted with a marvellous voice, but he worked on it like a
trouper all his life.
Here Not only functions as an adverbial in the first clause, and it is thus not
the first part of a correlative conjunction. Semantically, the combination not
only ... but is very similar to both ... and, for it has inclusive meaning and
emphasizes the bipartite nature of the stretch of speech it applies to. But this
emphasis is stronger than the one signalled by both ... and, and the state of
affairs referred to by the second clause, group or word is presented as
particularly noteworthy.
In closing this section we should point out that and and or are not in all cases
used as genuine coordinators. In a sentence like Try and find another, for
example, and is not used to coordinate two verbs but as an informal variant
of the infinitive marker to (which together with the verb and the pronoun
functions as a clausal object in Try to find another). And in an example like
Hang on a minute or two, or is not a genuine coordinator either but is used to
signal a rough temporal approximation. What we find here are special
idiomatic uses of and and or.
6.3. Subordination
6.3.1. Introduction
As argued in chapter 4, constituent structure does not reveal everything there
is to say about the relationship between constituents. Our sentence analysis
system plots fairly crude consist-of relations but does not really specify the
nature of the relationship between, say, a dependent article and a head noun,
or between a predicator and a direct object, except by marking the
constituents involved as precisely DEP, H, P and O, respectively. On the
contrary, dependents and heads, as well as predicators and objects, are often
placed at the same level in our analysis. While coordination is catered for
explicitly in our system by the compound unit (with its coordinators and
conjoints at the same level of analysis), subordination is often left implicit.
Occasionally subordination between two constituents can be shown by
placing them at different levels in our analysis. Thus, a subordinate clause is
analysed at a lower level than the main clause within which it is embedded
(see e.g. section 3.3.5). However, there is no obvious single configuration for
showing subordination between units within a group, or within a clause.
Stacking (see section 4.1) is no solution to this problem. By definition,
stacking is used to indicate that certain constituents are more closely related
than others (e.g. in a predicate or in a complex group head). But it does not
specify the nature of such closeness relations.
In the following sections we shall look at some of the main types of
subordination between constituents, all of which elude our sentence analysis
system.
6.3.2. Subordination at clause level

In many descriptions of English grammar some sentence constituents are
assumed to be subordinate to others. Very commonly objects, complements
and adverbials - but not subjects - are analysed as subordinate to the
predicator. In an example like She left the groceries on the kitchen table the
Subordination 179
noun group realizing the object and the preposition group realizing the
adverbial are thus considered subordinate to the verb realizing the predicator.
In syntactic analyses, the predicator and its subordinate constituents are
therefore often grouped together in a comprehensive predicate stack, the so-
called verb phrase. Furthermore, it is common practice to speak of verbal
complementation. This term refers to the way in which the action specified
by the verb can be complemented - or filled out - by an object as in kill the
enemy, a complement as in seem right, an adverbial as in live in Cambridge,
and so on. While the units realizing objects, complements and adverbiale are
thus regarded as subordinate to the verb, the unit realizing the subject can
hardly be considered coordinated with the verb. In coordination, typically,
one conjoint must be able to stand for the whole, and this is obviously not the
case in constructions like John snores. What we find here is neither a
dependent relation as in subordination (e.g. quite friendly) nor an independ-
ent relation as in coordination (e.g. tall and mighty) but an interdependent
relation. The unit realizing the subject is dependent on the unit realizing the
predicator and vice versa. Traditionally, this relation has been termed nexus
and characterized as more dynamic than dependent and independent
relations. While a group like the skating girls can be likened to a still, a
clause like The girls are skating can be likened to a film.
In some grammatical theories there is assumed to be subordination of a
different sort between the sentence constituents. Here the verb is regarded as
superordinate not only to the units realizing objects, complements and
adverbials but also to the unit realizing the subject. This type of analysis is
based on the assumption that the verb constitutes the pivotal part of the
sentence. It is characteristic of so-called valency grammar (the term Valency1
being taken over from chemistry, where it refers to the combining properties
of atoms), according to which the verb has one or more dependents, or
valents. A verb like DISAPPEAR takes only one dependent (a subject) and is
accordingly termed monovalent; KILL takes two (a subject and an object)
and is termed bivalent; and GIVE takes three (a subject, an indirect object
and a direct object) and is termed trivalent. This type of analysis is found in
other versions of dependency grammar as well and also in the theory of
functional grammar.
Accepting that the verb is the pivotal constituent of a clause (inviting
description in terms of valency) is not strictly incompatible with the view
that the actual relationship between subject and predicator is one of inter-
dependence and that the actual relationship between predicator and object (or
complement or adverbial) is one of complementation. This is the position
adopted in this grammar.
6.3.3. Subordination at group level

In groups we find three main types of subordination, all of which can be
illustrated by the following example:
(1) By the time he left the Dean's room the painful erosion of his self-respect
was almost complete.
In the painful erosion of his self-respect - in which the noun erosion
functions as head - there are three dependents, realized by the article the, the
adjective painful and the preposition group of his self-respect. The type of
subordination realized by the is determination, that realized by painful is
modification, and that realized by of his self-respect is complementation. In
the example, the signals that the noun group has definite reference, painful
adds descriptive meaning to the meaning of the head noun, and the preposi-
tion group of his self-respect completes this meaning, i.e. the action specified
by the deverbal noun erosion (derived from erode) is 'filled out1 by this
group. Let us look more closely at these three main types of subordination.
A) Determination. In English, determination can be separated fairly clearly
from modification and unproblematically from complementation, for while a
determiner always precedes the head of a group (typically a noun group), the
unit realizing a complementation always follows it (except in certain cases of
inversion). The words which function as determiners are placed in the
position before a premodifying adjective. They generally serve to signal what
kind of reference a noun group has, for example definite as in the girl and
indefinite as in a girl. The following forms are used as determiners (see
section 10.3): the articles (the, a/an), the demonstrative pronouns (this, that,
these, those), the possessive pronouns (my, your, etc.), genitives (Peter's,
boys', the shrewd politician's, etc.), the w/z-pronouns what, which, whose
(whether interrogative or relative) and a number of indefinite pronouns
(another, any, each, either, every, neither, no, some). In the following
sentence there are four instances of determination:
(2) If this lamentable speech was anything to go by, his statements must have
raised some hackles on the back benches.
B) Complementation. While determination applies mainly to noun groups,
the type of subordination termed complementation is more generally
applicable, not only to noun groups (e.g. need for expansion), but also to
adjective groups (immune to criticism), preposition groups (across the street)
and adverb groups (fortunately for me). There is a close relationship between
complementation at group level and complementation at clause level (see
section 6.3.2 above), as is borne out by constructions like We hesitate to
accept your offer (complementation at clause level), We are hesitant to_
Subordination 181
accept your offer (complementation in adjective group) and He noted our

hesitancy to accept your offer (complementation in noun group). Semant-
ically, complementation contrasts with determination by filling out the
meaning of the unit complemented. Thus to accept your offer fills out the
meaning of hesitant. Typically, if we leave the complementation of a unit
out, it can be assumed to be understood in the context, e.g. We are hesitant
can only be used appropriately if it is clear from the context with respect to
what we are hesitant.
While in most cases there is only one complement in a group, we
occasionally do find examples with two or more complements such as
responsible [to the School] [for keeping the class in order] and his donation
[of the money] [to the trust] [on Tuesday night]. Note that if the head word
is a nominalized verb (such as donation in the second example), each
complement represents a clause function in a corresponding clause (cf. He
donated the money to the trust on Tuesday night).
C) Modification. The third main type of subordination is called
modification. As pointed out in section 3.3.1, the nature of the relationship
between head and dependent is far from uniform. While a determiner
typically serves to signal what kind of reference a noun group has and a
complementation fills out the meaning specified by the word of which it is a
dependent, broadly speaking a modifier qualifies the meaning of its head
word, i.e. it attributes a property to the referent of this word. And unlike
determination, which is restricted to the position before the head, and
complementation, which is normally restricted to the position after the head,
modification occurs both before and after the head. In groups like depraved
tastes, freely available and only rarely there is premodification and in a
creature of habit, the woman I love and young to be a pilot there is
postmodification. Simultaneous pre- and postmodification is common as well
and can be exemplified by a group like the new police station that they built
last year. Unlike missing complementation, missing modification is not
implied. For example, by saying the woman instead of the woman I love, the
speaker does not imply / love but simply assumes that he has offered enough
information for the hearer to know whom he is talking about.
D) Other types of subordination. A possible fourth type of subordination in
groups is found in cases where two or more words make up a close unit for
the expression of a single concept, as, for example, in HOUSE RULES,
'DANCING MASTER (with primary stress on dancing, indicated by '*', as
distinct from "DANCING 'MASTER with primary stress on master), DOC-
TOR'S DEGREE (i.e. a particular kind of degree, not the degree attained by a
particular doctor). We consider such examples to be cases of complex
lexicalization or lexical compounding. Subordination is clearly present in

lexicalization of this kind: in e.g. HOUSE RULES, house is subordinate to
rules both semantically (because the expression refers to a certain kind of
rules, not a certain kind of house) and syntactically (since the word rules
rather than house dictates concord: Our house rules are most reasonable).
Arguably, there is complex lexicalization involving subordination in
complex predicators (phrasal verbs, prepositional verbs, etc.) such as GIVE
up (= 'yield' or 'surrender'), WAIT on (= 'serve'), TAKE place (= 'happen'),
etc. (cf. section 4.3). While subordination in complex lexicalization is 'to the
left' in noun groups, it is 'to the right' in verb groups, where e.g. up in GIVE
up must be considered the dependent if we want to preserve the status of the
group as a verb group. Complex lexicalization is present also in the so-called
complex prepositions: e.g. out o/(cf. into), because of, instead of, according
to, due to, in view of, in spite of, for (the) sake of, etc. But here internal head-
dependent relationships are more difficult to determine and are best left
unanalysed to indicate the idiomatic unity of the 'groups'.
An important subfunction in noun groups is quantification (see section
10.1.4). Quantification is syntactically less constrained than modification,
determination and complementation: it is not associated with a particular
kind of constituent. Quantification may affect the head noun directly (as in
the singular/plural number distinction: car/cars). At the same time however,
it may affect determiners (e.g. this car/these cars/a car/some cars) and
modifiers (e.g. numerous cars). A number of words are recognized as
'quantifiers': e.g. all, both, half, many,few, several, etc.; multipliers (double,
twice, three times, etc.); fractions (two-thirds, one-fifth, etc.), cardinal and
ordinal numbers (one, two, three, first, second, third, etc.). Such quantifiers
are clearly subordinate to the head noun but it is not entirely clear what kind
of subordination is involved: most often they are simply treated as a special
kind of determiner or as a special kind of modifier (for further discussion,
see section 10.3.3).
Finally, as we saw in section 3.3.2, it is by no means obvious that there is
subordination in preposition groups. Traditionally, in fact, a preposition is
assumed always to be accompanied by an overt complement, and in many
descriptions of English grammar there is accordingly said to be a relation of
interdependence between preposition and complement. In this book,
however, we take the view that the type of relation involved is dependence,
i.e. that the complement is subordinate to the preposition. We also retain the
traditional term 'complement' since the kind of relationship between
preposition and complement is complementation rather than modification or
determination: a prepositional complement 'fills out' the meaning of the
preposition (compare He helped an old lady across and He helped an old
Markers of clausal subordination \ 83
lady across the street}. As pointed out in section 3.3.2, in a preposition group
the preposition is the head by virtue of being the characterizing element of
the group without which the group would have been some other group. The
prepositional complement, which may be realized by any form type (word,
group, compound unit and (nonfinite) clause) without this affecting the status
of the preposition group as a whole, is the dependent. Another argument in
favour of this analysis is government those English pronouns which have
two case forms always select the objective case after a preposition. This can
be illustrated by examples like Give it to me. I've received a letter from them
and She is a woman with whom age doesn't count. Here the case of the pro-
noun realizing the complementation depends on the preceding preposition.
Subordination in groups will be further examined in the chapters on the
individual groups.
6.4. Markers of clausal subordination

6.4.1. Subordinating conjunctions
When clauses are linked together, subordination is usually indicated by
means of a connective word. One way in which this can be done is to use a
subordinating conjunction, i.e. a word which does not realize any of the
regular functions within the subclause (S, P, O, C, A) but constitutes a
linking element. The commonest subordinating conjunction is that:
(1) What is needed is that the two governments signal quite clearly that they are
looking for a new beginning.
(2) That she should do a thing like that is unbelievable.
(3) My wife is most anxious that you should dine with us one evening.
In some contexts that can be omitted. This is particularly common when the
clause it introduces realizes an object: We always knew 0 this sort ofthing
went on. But there may also be ellipsis where the subclause realizes an
extraposed subject (It is unbelievable 0 she should do a thing like that), a
subject complement (The main thing is 0 the audience should be curious) or
adjective complementation (I'm sure 0 this is just a misunderstanding). In
such cases, then, subordination is not overtly marked, but it can be recog-
nized by the possibility of inserting that.
In yes-no interrogative clauses subordination is signalled by the conjunc-
tions if ana whether:
(4) I wonder //he's got the qualifications for that job.
(5) It is not known at the moment if the intruder had intended to burgle the flat.
(6) Primary schools will be required to consider whether to offer sex education
lessons.
(7) I am uncertain whether we should go ahead as planned.
In English there is a fairly large class of conjunctions which are used to
introduce adverbial clauses:
(8) I know that the world has always been and will always remain as it is.
(9) I will write it if somebody wants it.
(10) The 80-year-old woman was attacked on her doorstep when she answered a
knock on her door.
(11) He wants Anglo-Catholics to be able to join the Roman Catholic Church
while retaining an Anglican liturgy and identity.
The class of subordinating conjunctions introducing adverbial clauses also
includes after, before, because, once, since, (although, till, until, -whenever,
where, whereas, whereupon and whereever. In British English, furthermore,
whilst is used as a variant of while. Attention should also be drawn to lest
and like, which are particularly common in American English, and which in
British English are formal and informal respectively.
As in the case of that, //and whether, we assume that the subordinator
introducing an adverbial subclause does not realize a regular clause function
within the clause it introduces but is a special peripheral element (SUBxonj).
We thus analyse an example like / will write it if somebody wants it in the
following way:
STA
decl
pro
I will write it SUB

conj pro pro
if somebody wants it
At first blush it is less obvious that the subordinators introducing adverbial

clauses should be analysed as peripheral in the clause they introduce than in
the case of that and of ι/and whether in interrogative subclauses. While that
is semantically empty and if and whether have highly general meaning,
subordinators like after, because, before, once, since, when(ever) and where
are less evidently function words. It is tempting, therefore, to analyse them as
realizations of an adverbial in the (adverbial) subclause they introduce rather
Markers of clausal subordination 185
than simply as subordinating conjunctions. However, if we consider sen-

tences like the following more carefully it appears that this analysis is not
tenable:
(12) She got depressed because he left.
(13) She got depressed after he left.
(14) She got depressed whenever he left.
In these examples, it is surely not the situation referred to by the subclause
(i.e. 'his leaving') that is explained (by because) or located temporally (by
after and whenever) but the situation referred to by the matrix clause ('her
getting depressed'). In other words the 'adverbial force' of because, after and
•whenever is directed towards the matrix predicator rather than the predicator
in the subclause. This means that the content of such words ensures an
adverbial interpretation of the subclause as a whole in relation to the matrix.
There is no sense in which they can be said to function like adverbials within
the subclause. The function they serve here is merely to connect the clause as
a subclause in (adverbial) relation to the matrix and therefore they should be
analysed as SUB:conj just like that, if ana whether. The difference between
all these conjunctions is simply the degree of semantic specificity with which
the subclause is connected with the matrix.
6.4.2. Interrogative and relative pronouns

Subordination may also be marked by interrogative or relative pronouns. In
this case the connective word is integrated in the subclause and realizes an
object, a complement, an adverbial or part of such a constituent. The
following examples illustrate interrogative pronouns in subclauses:
(1) I've discovered who my real friends are. (C)
(2) The police have no idea whose car it is. (part of C)
(3) Tell me which you prefer. (O)
(4) You're probably wondering for what specific reason I've resigned, (part of A)
Here subordination is signalled by the combination of the interrogative
pronoun and the non-inverted S P order in the subclause. In the correspond-
ing main clauses there is subject-operator inversion (e.g. Whose car is it? and
Which do you prefer?). If the interrogative pronoun realizes the subject in the
subclause, there is no formal marker of subordination, compare:
(5a) I asked her who did it.
(5b) Who did it?
In relative clauses the subordinating pronoun also realizes a syntactic
function within the clause it links to the preceding noun, pronoun or clause.
Besides who, whom, which and whose the class of relatives includes the
pronoun that and 0 (for missing relative):
(6) He will be missed by all who knew him. (S)
(7) Ian, whom I trusted with the money, has disappeared. (O)
(8) She talked to the students whose parents had complained. (DEP in S:g)
(9) This is the book 0/that/which I talked to Jill about. (DEP in A:g)
In examples like / did not like what I saw and What I want is to get on with
the resolutions, what is traditionally analysed as an 'independent relative
pronoun' (cf. / did not like that which I saw) because it has no antecedent.
Like the other relative pronouns it signals subordination but realizes a
function in the subclause other than SUB.
In tree diagram analysis, subordination of interrogative or relative clauses
is represented in the following way:
STA
cl
S P Ο
pro v cl
wonder _P Ο
pro g pro adv
who 's kissing her now
STA
decl
S_ P A.
pro g g
He is missed H DEP
prep
by
all
who knew him

Markers of clausal subordination \ 87
In other words, interrogative and relative pronouns are assigned functions

like S, O, DEP, etc. rather than SUB, which is reserved for elements with a
purely subordinating function.
In an example like The village in which the murder was committed has
been thrown into a turmoil the adverbial of the subclause is realized by a
preposition group which has a relative pronoun as its complementation. Here
it would also be possible to use where instead of in which. Some gram-
marians therefore classify this word as a relative adverb and assume that it
realizes an adverbial in the same way that in which does. Other words
analysed as relative adverbs are when, why and how which may be
considered replacements of at which, for which and the way in which
respectively in examples like That was the time when he entered the house,
That is the reason why I called you and That is how she died. If this analysis
is accepted, it would not be unnatural to recognize interrogative adverbs as
well. In e.g. / wonder when/where/why the meeting is taking place the wh-
words could be considered replacements of at what time, at what place and
for what reason, respectively. In e.g. Do you know how I can get to Ely,
similarly, how could be considered a replacement of in what way or by what
means and be assumed to realize an adverbial in the subclause. Note that
unlike when as a subordinating conjunction (as in She got depressed when he
left, cf. our discussion in section 6.4.1 above), when as an interrogative or
relative adverb functions adverbially in the subclause. Thus in examples like
That was the time when he entered the house and / wonder when the meeting
is taking place, the meaning of when clearly relates to the situation expressed
by the predicator in the subclause. In this connection it is interesting to note
the difference between the following two (sets of) examples:
(10) I wonder ifwhether he's got the qualifications for this job.
(11) I wonder when/where/why the meeting is taking place.
If the subclauses in these examples are turned into main clauses we get:
(10') Has he got the qualifications for this job?
(1Γ) When/where/why is the meeting taking place?
The fact that //and whether are dropped but when, where and why are
retained in main clauses indicates that the former are clearly to be analysed
as SUB:conj in the corresponding subclause while the latter are clearly to be
analysed as A:adv.
6.4.3. Complex subordinating conjunctions

Subordination can also be indicated by a complex subordinating conjunction,
i.e. a linker consisting of more words than one:
(1) You're as pale and drawn as //you just got out of a sickbed.
(2) He smiles as though he knows the final truth and I can't stand that smirk.
(3) The apartment isn't mine, and even if'it was, I couldn't pay for it.
(4) The police play with us like cats with mice. But so long as we remain on this
side of the bars, we don't feel like being alone.
(5) As soon as he leaves the house, I start looking up at the ceiling for a hook.
(6) The diagnosis is made harder if a child's gifts go unnoticed for a while, so
that frustration reduces his or her academic performance.
Other complex subordinators which are common in English include in case,
in (order) that, provided (that) and on condition (that). Like the ones
illustrated by the examples above, we choose to interpret these as (complex)
conjunctions, i.e. as peripheral elements in the subclause.
6.4.4. Correlative subordinators

According to some grammarians, English has not only correlative
coordinators (both ... and, either ... or and neither ... nor, see section 6.2.4)
but also correlative subordinators. The most natural candidates for
membership of such a class are if... or and whether ... or. In a sentence like
We must decide if/whether I should call him or you should write him the
object is realized by two subclauses the subordinate status of which might be
said to be signalled by //... or or whether ... or as a unit. In favour of
analysing whether ... or as a correlative coordinator it might be argued that
when whether introduces a conditional-concessive clause, it is obligatorily
accompanied by or and another such clause. This can be illustrated by an
example like I'm going whether you wish to come with me or prefer to stay
here from which the last five words cannot be omitted. The argument cannot
be extended to all cases, though. For one thing, if cannot introduce a
conditional-concessive clause. Further, when whether or if introduces an
interrogative subclause, it need not be followed by or and another such
subclause. From the first example given in this paragraph we could omit or
you should write him without making the sentence ungrammatical. In such
cases it would be natural to assume that subordination is signalled by
whether or z/exclusively and that or is an independent coordinating
conjunction, i.e. that correlative subordination is not involved. When whether
... or and i f . . . or are used, it is typically stacks of constituents which are
linked together by or. If we employ the stack convention and assume that or
realizes a coordinator, we can analyse an example like / wonder if/whether
she loves me or hates me in the following way:
Markers of clausal subordination 189
QUE
cl
S_ P Ο
pro v cl
wonder SUB S X_
conj pro cu
if she CJT
τCO CJT
/whether conj
P Ο or £
v pro ν pro
loves me hates me
In favour of analysing if and whether as independent of or, it could be argued

that they can also be followed by structures which are linked by the
coordinating conjunctions and and but: I wonder if/whether she loves me and
hates me at the same time /Iwonder if/whether she loves me but fears me.
A subordinating conjunction often correlates with a following conjunct,
though never obligatorily:
0) If a child's gifts go unnoticed, then the child will get frustrated and perform
less well academically.
(2) Although my Volvo is from 1985, it is nevertheless very reliable.
In examples of this type there is only one subordinator (the conjunction), but
the meaning of the word realizing it is reinforced by the conjunct in the
following clause.
6.4.5. Other markers of subordination

Subordination can also be signalled by nonfinite verb forms or by zero-
realization of a predicator:
(1) I watched her cross the street.
(2) It is refreshing to describe α major American company that is successful.
(3) Anyone doing anything in computers faces the formidable task of competing
with IBM.
(4) The objects of all that attention are elusive submarines, widely believed to be
Russian.
(5) Two people, both of them from Rainham, Kent, escaped injury after their
light aircraft crash-landed in a cricket ground.
Furthermore, subordination can be signalled prosodically. This is the case in
comment clauses, in which there is often no subordinating conjunction, and
reporting clauses. In examples like We have no white coats here, you'll have
noticed and We have no white coats here, she added the subordinate status of
the second clause appears from the fact that it is pronounced without pitch
prominence (i.e. on a low tone level) and with weak stress.
In colloquial spoken English we may encounter examples like the follow-
ing (cf. Hopper & Traugott 1993: 173):
(6) You keep smoking those cigarettes, you're gonna start coughing again. (= 'If
you keep smoking those cigarettes,...')
(7) That gay just walked out the store, he reminds me of the photo in the post-
office window. (= "That guy who just walked out the store,...')
In the first example, the italicized clause is a conditional subclause and in the
second example the italicized clause is a relative clause with guy as the
antecedent. In both cases, the subordination of these clauses is indicated
prosodically through rising intonation. With falling intonation they would be
interpreted as main clauses.
Finally, attention should be drawn to two minor markers of subordination,
namely inversion in conditional clauses (see section 5.3.3) and the use of
what and how in exclamatory subclauses:
(8) Had he reached the balcony a moment earlier, he would have heard the rattle
of the chain and the click of the bolts.
(9) It surprised me what a boring lecturer he is.
(10) I'm impressed how confidently you work the machine.
7. The simple sentence
7.1. Introduction: simple and complex sentences

Sentences are either simple or complex. A sentence is simple if it consists of
one clause only:
(1) The papers blew from my desk.
(2) Marion is such a nice person.
A sentence is complex if it contains one or more subclauses:
(3) If we are attacked we will defend ourselves.
(4) Police believe that the shooter was a professional.
In this chapter we take a closer look at a number of issues relating to the
simple sentence and its internal organization: referential properties (section
7.2), participant roles of clause functions (section 7.3), voice (section 7.4),
polarity (section 7.5), subject-predicator concord (section 7.6) and other
types of external concord (section 7.7). Much of the discussion here applies
also to the complex sentence, which is examined in chapter 8. The reasons
for dealing with the two types of sentence in different chapters are that
simple sentences provide a convenient pedagogical descriptive basis and that
complex sentences display a number of unique features which require
separate treatment.
7.2. Referential properties: situations

Language allows us to talk about all the goings-on, dealings, emotions,
perceptions, attitudes, etc. that are part of everyday human lives. With lan-
guage we also identify things, and we classify, characterize and relate them.
In this book, the term 'situation' is used as a cover term for the many different
meanings that sentences express. For example, by saying Jack fixed the old
motorbike, the speaker informs the hearer of an instance of 'fixing1
(expressed by fixed) in which there are two participants: someone who does
the fixing (Jack) and something that is affected by this, i.e. something that
gets fixed (the old motorbike). These components make up a total meaning, a
proposition expressing 'the situation of Jack fixing the old motorbike'.
7.2.1. Actionality: dynamic vs. stative situations

As human beings we conceive of the world and of all the situations taking
place or existing in the world in terms of differences and similarities. For
192 The simple sentence
example, the situation expressed by a sentence like The old man painted the
wall shares a number of features with the situation of 'Jack fixing the old
motorbike1: in both cases there is a participant doing something to another
participant with a certain result (the motorbike gets fixed and the wall gets
painted) and both situations endure over time before they reach a natural
endpoint. For this reason it is tempting to say that they belong, at some level
of classification, to the same type of situation despite the fact that they
obviously differ with respect to all the more specific characteristics. Turning
now to an example like Ottawa is the capital of Canada, we find that here
there is no participant doing something to another participant. Nor is there
any natural endpoint or result of an activity. Rather, the example expresses a
relation between two (referentially identical) 'participants': one (Ottawa) is
characterized in terms of the other (the capital of Canada). By looking at a
large number of sentences, we discover that it is possible to classify the
situations they express (real-world, hypothetical or fictional) in situation
types on the basis of our conception of their differences and similarities.
The study of situations expressed by sentences is called the study of
actionality. The different types of situation that we recognize are described in
terms of the action category. Like other categories (such as tense, aspect and
mood), the action category is very controversial: few linguists agree on
exactly what situation types should be recognized, what they should be
called and how they should be defined. In the following, we shall restrict
ourselves to certain fundamental distinctions which are relevant to English
grammar (such as the active/passive distinction and the use of the pro-
gressive verb form).
It is important to note that what we are doing here is to typologize
situations referred to by sentences, not sentences referring to situations. Nor
should the typology of situations offered below be confused with a typology
of verbs. How exactly verbs are used in language expressions referring to
situations is a topic dealt with in chapter 9 below.
The primary actional distinction is that between dynamic and stative. A
dynamic situation requires a continual input of energy and typically involves
change while a stative situation requires no input of energy and remains the
same. A dynamic situation happens or takes place while a state exists or is
true of someone or something. The situations of 'Jack fixing the old motor-
bike' and The old man painting the wall' are both dynamic whereas the
situation of Ottawa being the capital of Canada' is stative.
Referential properties: situations 193
7.2.2. Subtypes of dynamic situations

There is, strictly speaking, no end to the number of subtypes of dynamic and
stative situations that may be identified for the action category. However,
some subtypes are more important than others in the description of the
English verb. Let us look first at the most important dynamic ones:
Punctual situations have little or no extension in time and hence no internal
structure. The situation referred to by e.g. He suddenly switched from
Spanish to English is punctual in this sense. Other examples: She hit me hard
on the nose / Stephen dropped the gun. Beginnings and endings are also
punctual situations, e.g. He started running and He stopped talking.
Telic situations are conceived of as durative (i.e. as having extension over
time) leading up to and including a natural terminal point beyond which no
further progression is possible or relevant, and without which the situation is
not fully completed. The situations of 'Jack fixing the old motorbike' and
'The old man painting the wall1 (described above), as well as the situation
referred to by e.g. The student wrote an article about the split infinitive, are
telic. The term 'telic' is based on Greek telos 'goal, end1.
Directed situations progress towards a natural terminal point but do not in
fact include this point. The situation referred to by e.g. Sally was building a
small garden shed is directed in this sense: the Tmilding' activity is directed
towards a point of completion but this point is outside the situation expressed
by the sentence, not inside as in the case of telic situations. Other examples:
The girl was catching up with the rest of us /He slowly approached the door
/She tried to solve the problem.
Self-contained situations are durative situations conceived of as not having,
or being directed towards, any natural point of completion. In other words,
they may be terminated at any time without this affecting their completeness.
Self-contained situations are the least dynamic and the most stative of the
dynamic situation types. Examples: James and George were sailing along
the coast/ We were celebrating Stephanie's birthday at my uncle's place / We
discussed their predicament at the last meeting.
Iterative situations consist of a number of identical, or similar, consecutively
realized subsituations. The situation referred to by Jack was knocking at
Sally's door is thus complex: it consists of a number of knocks. Other
examples: Someone was tapping me on the shoulder / The telephone was
ringing /Roger kept calling me.
To fully understand the nature of actional subtypes, it is essential to make a
distinction between 'the real world' and 'the world as conceived by the
speaker', i.e. between objective facts and the subjective way that we think of
these objective facts or choose to think of them in a particular context.
Actional distinctions are to do with our conceptualization of the world rather
than with the world itself. Thus, for example, the situation expressed by an
example like The bus stopped for a red light is probably best classified as
punctual from the point of view of the action category (because we tend to
conceive of'stopping' as a punctual situation), but, strictly speaking, it is not
punctual in the real world (because here it involves braking and slowing
down to a complete halt). Consider also:
(1) Jack was reading the report when I got back last night.
(2) Jack was reading when I got back last night.
Whether we choose to refer to a situation of'reading' as directed (as in (1)) or
as self-contained (as in (2)) may have very little to do with the objective
reality of the situation, which may well be the same. What matters is what
the speaker wishes to make of this situation in his message to the listener.
7.2.3. Subtypes of states

Stative situations, which are almost always relations of some sort, can be
subclassified in this way:
Intensive relations. An intensive relation involves either a description of an
entity in terms of another (for identificatory or classificatory purposes) or an
assignment of a property to an entity (characterization). The situations
referred to by Ottawa is the capital of Canada and Victoria is beautiful are
both intensive relations. Other examples: They were in high spirits /Joan is
the small girl talking to Bob over there/She seemed very unhappy,
Extensive relations. An extensive relation is a physical state, condition,
location, position or possession obtaining for an entity. The situation referred
to by e.g. The village lies in a dark valley is an extensive relation in which
the village is specified with respect to location. The situation referred to by
an example like Ildiko has a red Mercedes Benz is an extensive relation
between a person and a thing in terms of possession. Other examples: This
factory belongs to Mr Hardcastle / The key is in the top drawer / This box
contains all his private papers.
Attitudes. An attitude is a psychological state (opinion, belief, love, hatred,
liking, need, knowledge, supposition, etc.). The situation referred to by e.g.
George believes in God is an attitude in this sense. Other examples: 7
appreciate all her help /She hates all the fuss /Everybody knows the truth
about Jim /Nobody likes me / They wanted it all.
Referential properties: situations 195
Perceptions. A perception is a sense relation (visual, auditory, etc.). The

situation referred to by e.g. / saw her clearly from my bedroom is a
perception, more specifically a visual sense relation. Other examples:
Charles heard her cries in the distance/She sensed a certain uneasiness on
his part /I felt her damp hair against my skin.
Habits. A habit is the product of a (dynamic or stative) situation occurring so
regularly that it is conceived of as characteristic of someone or something.
The situations referred to by e.g. John teaches linguistics and Sally smokes
are not dynamic instances of teaching and smoking, respectively (though
such instances are implied), but rather characterizations of John and Sally, cf.
John is a teacher of linguistics and Sally is a smoker. The term 'habit1 should
be understood in a broad sense as including not only personal habits (like
Roger plays the guitar, Victoria would call me every day before breakfast)
but also 'universal truths' (e.g. The sun rises in the east) and 'ability' (e.g.
Vera speaks Italian). Like iterativity, habituality is situationally complex,
often implying a dynamic input.
Many verbs have a strong potential with respect to the specific values of the
action category: thus verbs like HIT, DROP, CRASH, OPEN, SWITCH,
SNATCH, START, STOP, etc. are often used as predicators in sentences
referring to punctual situations and verbs like BELIEVE, SUPPOSE, ASSUME,
PRESUME, KNOW, etc. are often used as predicators in sentences referring to
attitudes. This has led many grammarians to talk about verb classes such as
e.g. 'punctual verbs' and 'attitudinal verbs'. Such terminology is misleading. It
is important to realize that there is seldom a one-to-one relationship between
lexical verb and type of situation referred to. As we saw in the last section,
the presence of a direct object after a verb like READ may turn a self-
contained situation (He was reading) into a directed one (He was reading the
report). Many verbs which regularly appear in sentences referring to
dynamic situations can be used statively about habits; cf. Roger was playing
the guitar (self-contained situation) vs. Roger plays the guitar (habit). And
even verbs with a distinct potential for expressing punctual or otherwise
dynamic situations can be used to express extensive relations, as in:
(1) The cliffs dropped into the endless yellow sabre of the Chesil Bank.
(2) The path climbed and curved slightly inward beside an ivy-grown stone wall.
A little beyond them the real cliff plunged down to the beach.
Conversely, verbs with a strong stative potential like KNOW may occasion-
ally be used dynamically, as in e.g. Roger at once knew (- 'realized1,) what
was wrong. To cut a long story short: the action category allows us to
classify situations, not verbs. Obviously the choice of a lexical verb is
important for the type of situation referred to. But so are a number of other
considerations: inflectional form (e.g. the progressive/nonprogressive
distinction and the present/past distinction), syntactic relations (e.g. the
presence or absence of a direct object) and the extralinguistic context in
which the sentence is used.
7.3. Participant roles

7.3.1. Introduction: general roles
The study of the number and kinds of participants that may be associated
with verbs is called the study of valency. Like the category of action, valency
is a controversial subject: there is no generally accepted terminology, and
few linguists agree on how many participant roles should be identified and
exactly how they should be defined. Strictly speaking, it is possible to argue
that every verb has its own participant roles: KILL (killer, victim); OWE
(creditor, debt, debtor); etc. However, as with the action category, we
examine a number of roles that are important to the grammar of English.
Together with action, valency may contribute to a better understanding of the
semantics of the basic sentence patterns.
First we draw a distinction between those participant roles which are
involved in dynamic situations and those which are involved in stative
situations. The typical dynamic situation involves a 'DOER' and a 'DONE-
TO', i.e. someone /something bringing the situation about and someone or
something passively affected by the situation. In Jack fixed the old
motorbike, the subject Jack is the DOER and object the old motorbike is the
DONE-TO (it gets fixed). In The landslide killed the old man, similarly, the
landslide is the DOER and the old man is the DONE-TO (he gets killed). In
addition to these very general primary participants, there are a number of
relevant EXTRAS. For example, in Roger teased the rat with a stick, the
subject Roger is the DOER, the object the rat is the DONE-TO (it gets
teased) and the adverbial with a stick is an EXTRA, indicating more
specifically the instrument with which Roger brings about the situation.
By contrast, the typical stative situation, being a relation rather than a
going-on, involves a 'SPECIFIER' and a 'SPECIFIED'. A SPECIFIER
determines the nature of the state (relation) in conjunction with the
predicator, and a SPECIFIED is someone/something for whom/which the
state exists or is true. In Jack is in London, Jack is SPECIFIED with respect
to the locational SPECIFIER in London. In The girl is exceptionally clever,
the girl is SPECIFIED with respect to the qualitative SPECIFIER
exceptionally clever. Like dynamic situations, states may involve various
Participant roles 197
EXTRAS. For example, in Jack was in London last week, the adverbial last
week is an EXTRA providing a temporal restriction of the extensive relation.
The very general roles proposed here are syntactically based in the
following sense. In active declarative sentences, the DOER or the
SPECIFIED always occupies subject position, the DONE-TO always
occupies object position and the SPECIFIER always occupies object or
complement position. Whatever functions as a subject in an active
declarative sentence expressing a dynamic situation is thus conceived of as
the DOER; whatever appears as the object in such sentences is conceived of
as the DONE-TO; and so forth. Thus, even in a problematic example like
John received the telegram, John is the DOER of the situation of receiving,
and the telegram is the DONE-TO. In other words, our general participant
roles reflect the semantic functions conceptually assigned at a very general
level to the primary syntactic functions.
7.3.2. Specific roles

It is useful to operate also with a central system of specific roles based on the
following two distinctions: dynamic vs. stative and volitional vs. non-
volitional. The possible combinations of the values in these distinctions yield
the following four more specific roles:
AGENT represents the volitional (typically human) instigator of a dynamic
situation (cf. Jackßxed the old motorbike).
CAUSE represents the non-volitional (typically non-human) entity bringing
about a dynamic situation (cf. The landslide killed the old man). Note that an
example like Sally ruined my marriage is ambiguous, or simply vague,
between an AGENT reading and a CAUSE reading of the subject noun.
CONTROLLER represents the volitional (typically human) participant for
whom a state obtains for so long as the controller keeps it that way (e.g. John
keeps a gun in the cupboard / Sally wants some ice cream / Roger is in
London /Mick Jones is a university professor / John teaches linguistics).
HOLDER represents the non-volitional (typically but not inevitably non-
human) participant for whom the state obtains (e.g. The village lies in a dark
valley / The jar contained some milk / Victoria is beautiful / The sun rises in
the east).
Schematically the central system of specific roles looks like this:
Dynamic Stative
Volitional AGENT CONTROLLER
Non-volitional CAUSE HOLDER
To this central system of specific roles we must add an important default

role:
AFFECTED represents people or entities crucially involved in, or affected
by, a dynamic situation (e.g. Jack fixed the old motorbike / The old man
painted the wall / Somebody had beat him up) or passively forming part of
the state description in extensive relations, attitudes, perceptions and habits
(e.g. John keeps a gun in the cupboard /Sally wants some icecream /Isaw
her clearly from my bedroom /Roger collects stamps).
Two other specific roles are recognized by most grammarians:
BENEFICIARY represents people or entities for whose sake the dynamic
situation is brought about (e.g. Roger bought Sally an expensive necklace /
Mother baked US. a chocolate cake).
INSTRUMENT represents entities or means (typically non-human) used to
bring about a dynamic situation (e.g. Roger peeled the potatoes with his
pocket-knife /Sally travelled by train).
In addition, it is often useful to extend the notion of'participant role* to more
general semantic meanings like the following:
ATTRIBUTE represents three stative subroles: characterization (as in
Victoria is beautiful), identification (as in Bill Clinton is the fellow in the
corner) and classification (as in Mick Jones is a university professor).
RESULT represents an entity created by the situation (as in He dug a hole)
or a change of state (She became a raving lunatic /He got very upset).
PLACE represents a variety of spatial concepts, such as goal or 'place to' (/
went to Rome), source or 'place from' (He left the mansion), 'path' (She
moved along the corridor), location (Jack was in London), etc.
TIME represents a variety of temporal concepts, e.g. temporal location (Jack
was in London last year), duration (He read for several hours), 'time as a
resource' (Time is running out / We spent too much time on the project), etc.
CIRCUMSTANCE represents a variety of more specific meanings express-
ing the background, circumstance, setting etc. of the dynamic situation or
some sentence-external relation (as in He assembled the model with great
care / Quite honestly. I didn't even kiss her goodbye / On top of all this, she
left her children / To my surprise, the jar contained milk only).
7.3.3. Additional points

Note first that participant roles are sometimes ambiguous or vague. In
examples like He fell to the ground, the subject could be interpreted as either
DOER AGENT or DOER AFFECTED depending on whether or not the
subject falls intentionally.
Secondly, participant roles may change according to the nature of the state
specified by the lexical verb even if the reality of the situation remains more
or less the same, which indicates that participant roles are often a question of
the speaker's presentation of situations; compare:
(la) [Mr wilson]CONTROLLER owns [this house]AFFECTED
(Ib) [This house]HOLDER belongs to [Mr Wilson]*™*™0.
In the first example, Mr Wilson is the SPECIFIED CONTROLLER of the
'state of owning' and this house is SPECIFIER AFFECTED. In the second
example, the same reality is presented differently: with This house as the
SPECIFIED HOLDER of the 'state of belonging to' and with Mr. Wilson as
SPECIFIER AFFECTED.
It is important to emphasize that there are many alternative ways of
handling the question of participant roles in English. We have tried not to
make our system too fine-grained. Thus, for example we do not want to
reflect the different degrees of 'affectedness' in units to which AFFECTED
applies (cf. e.g. Roger hurt Sally and Roger knows Sally}. Nor do we
distinguish between participants truly benefiting from a situation (as in
Roger bought Sally an expensive necklace) and e.g. Victims' (as in /'// give
them hell and She told us_ a pack of lies): in both cases we operate with the
label BENEFICIARY.
On the other hand, unlike many other systems, we operate with two levels
of role participation: the very general conceptual level associated with
syntactic functions (DOER, DONE-TO, etc.) and the more specific level
reflecting our knowledge of the nature of referents and their involvement in
situations (AGENT, BENEFICIARY, INSTRUMENT, etc.).
This characteristic allows us to handle many cases of abstract and meta-
phorical use of language. Consider, for example, the following sentences (cf.
Lakoff and Johnson 1980: 33ff):
(2) John attacked Sally.
(3) His theory attacked the very notion of transformations.
(4) Inflation attacked the foundation of our economy.
In all three cases, the direct object can be analysed unproblematically as

AFFECTED irrespective of the various degrees of concreteness. The subject,
however, is to be analysed very differently in the three examples: in (2) it is
AGENT; in (3) it is INSTRUMENT (cf. With his theory he attacked the very
notion of transformations); in (4) it may be interpreted variously as the
CAUSE of the attack on the foundation of our economy, or as the RESULT
of certain developments in our society which may affect the foundation of
our economy, or as a CIRCUMSTANCE threatening our economy. In any
case, the three subjects represent different degrees of abstractness: John is
very concrete; His theory is more abstract but probably tied to some written
exposition (e.g. a book or an article); and Inflation is - to most people - a
very abstract concept, which may explain our difficulty in analysing it in
precise participant terms. At the same time, however, all three subjects (even
inflation) are DOERs at the more general level of analysis. A DOER is
typically an AGENT (as in example (2)) but all sorts of other participants
may serve as DOERs (as in examples (3) and (4)). The more a participant
departs from our prototypical AGENT DOER by not being human, volitional
and concrete, the more we feel that language is used metaphorically: theory
and inflation are presented as if they were AGENTS. By thus relating the
two levels of role participation, we may capture the more elusive examples
of abstract and metaphorical use of language.
7.3.4. Formal links

In this section we briefly examine some of the ways in which valency can be
usefully integrated in the grammatical description of English:
A) The S P pattern. There are in this pattern few restrictions on the role of
the subject:
(1) The architects of this affair apologised. (AGENT)
(2) The boat capsized. (AFFECTED)
(3) All the guests stayed on. (CONTROLLER)
(4) Little of the original fortress remains. (HOLDER)
Note that there is often a correspondence between S P constructions like
Suddenly the rock moved and S P O constructions like Someone moved the
rock. In both cases, the rock is AFFECTED. The difference is that in the S P
construction the rock is presented as the DOER whereas in the S P O con-
struction, it is presented as the DONE-TO. In the passive counterpart to the S
Ρ Ο construction, The rock was moved, the noun group the rock is made the
syntactic subject as in The rock moved, but unlike this active S P con-
struction, it retains its analysis as DONE-TO. Similar trios of constructions
are found with verbs like OPEN, CLOSE, SCAN, LOCK, etc. (e.g. The gate
opened / Someone opened the gate / The gate was opened).
B) The S P A pattern. Syntactically there is a grey zone area between the S
P pattern and the S P A pattern because it is not always easy to determine
whether an adverbial is optional (as in She was sleeping on the floor} or
obligatory (as in She was in London). And the S P A pattern shows the same
range of semantic subject roles as the S P pattern. But while many different
kinds of adverbial may be included optionally in the S P pattern (as in e.g.
She was sleeping when I got back / heavily / on the floor), the adverbial in
the S P A pattern is more restricted. Typically we find PLACE adverbiale:
(5) The Ford went into the East Sector.
(6) Turkey will also stay at arm 's length.
But also adverbials expressing TIME and ATTRIBUTE are possible:
(7) His death came at the worst possible time.
(8) She was in high spirits.
C) The S P C pattern. When examples of this pattern refer to stative
situations we typically get HOLDER or CONTROLLER subjects and
ATTRIBUTE complements (expressing characterization, classification,
identification, or a combination of these):
(9) The marriage appeared happy.
(10) The banana plantation is a reservoir of rats.
(11) Schmeichel is the goalkeeper.
In dynamic instances of the S P C pattern we typically get AFFECTED
subjects and RESULT complements:
(12) Marital breakdown became a commonplace consequence of unemployment.
(13) She got pretty mad at me.
In dynamic examples like the following (discussed by Radden 1989: 459),
the subject receives double role analysis as the result of monovalent
intransitive verbs being used as divalent copula verbs:
(14) He was born a slave.
(15) He returned a new man.
In (14) the subject is both AFFECTED (in relation to the basically
intransitive predicator) and HOLDER (in relation to the imposed
complement ATTRIBUTE). In (15) the subject is both AGENT and
HOLDER (or CONTROLLER). Note that there is a temporal link between
the two role relations, cf. the following paraphrases:
(141) When he was born, he was (already) a slave.

(15') When he returned, he was a new man.
D) The S P O pattern. This pattern typically consists of AGENT subjects
and AFFECTED objects:
(16) This guy killed many people.
(17) Police searched the MP's office.
But it is also the pattern used for the expression of many states with
HOLDER or CONTROLLER subjects:
(18) Two towers support the middle part of the bridge.
(19) Roger plays the guitar.
With sentences referring to dynamic situations there is a lot of variation,
especially in the case of the subject:
(20) [He]AGENT dug [a big hole]^1^
(21) [The report]INSTRUMENT proved [his guilt]AFFECTED.
(22) [The wind]CAUSE broke [all the windows]AFFECTED.
In the following examples (from Radden 1989: 428) we find objects serving
two semantic functions as the result of primarily intransitive verbs being
used transitively:
(23) The guard marched the prisoners (to the camp).
(24) The lady is walking her spaniel (in the park).
While the subject in such examples is clearly AGENT (sometimes referred to
as the 'primary agent1) the object is both AFFECTED and AGENT
(sometimes referred to as 'secondary agent').
E) The S P O O pattern. The subject is here typically AGENT or CAUSE,
the indirect object BENEFICIARY and the direct object AFFECTED:
(25) He may have told the Russians everything.
(26) The book gave left-wingers a lot of inspiration.
A distinction must be made between BENEFICIARIES actually participating
in the situation as DONE-TO on a par with the AFFECTED direct objects (as
in (25) and (26)) and BENEFICIARIES not really participating in the
situation but having the general role of EXTRAS, as in:
(27) My mother baked us a chocolate cake.
(28) Roger ordered Stephen a new radio.
F) The S P O A pattern. In this pattern the subject is typically AGENT, the
object is AFFECTED and the adverbial is PLACE:
(29) The robbers put her in the boot of their car.

(30) She placed her grammar books on the topmost shelf.
But there is some variation, as we see in the following examples:
(31) [Mr Major]AGENT put [ÂFFECTED [differently]ciRCUMSTANCE
(32) [He]AGENT offered [a beer]^1*™15 [to „^BENEFICIARY
When the adverbial has the semantic function of ATTRIBUTE, the object
receives double analysis:
(33) [Clinton]AGENT described [theCase]AFFECTEI>fHOLDER [as serious]ATTRIBUTE
Here the case is affected in relation to the predicator but at the same time
there is an intensive relation between the case and as serious. The analysis of
the case as also HOLDER and of as serious as ATTRIBUTE captures this
intensive relation.
G) The S P O C pattern. In this pattern there is a secondary, intensive
relation between object and complement and this calls for a double analysis
of the object in terms of role participation. When the main situation
expressed is dynamic, the subject is prototypically AGENT or CAUSE, the
object is twice AFFECTED and the complement is RESULT:
(34) [Tyson]AGENT knocked [Bruno]AFFECTEIHAFFECTED [unconscious]
(35) [He]AGENT pronounced [US]AFFECTEIKAFFECTED [man ^ ^
The analysis of the object as AFFECTED+AFFECTED reflects the fact that
it is affected both in relation to the situation expressed by P (in e.g. Tyson
knocked Bruno unconscious, 'something happened to Bruno') and in relation
to the implied change of state ('Bruno became unconscious').
In some dynamic cases, the complement is more loosely attached to the
proposition than in the examples above (in the sense that the sentences make
perfectly good sense even if the complement is left out):
(36) [Roger]AGENT painted [the waiyAFFEcreiHAFFEcreD [blue]RESULT
Before concluding this section on formal links, let us briefly examine some
additional points in relation to role participation.
H) Optional adverbials. Optional adverbials may express a wide range of
semantic roles:
(37) The bike was stolen [by the neighbour's boy]AGENT.
(38) She left early [because of her sick mother] CAUSE
(39) Roger broke the window [with a hammer]INSTRUMENT
(40) I talked to her [for several hours]TIME.
(41) [In Paris]PLACE I told her my little secret.
To these can be added the default role CIRCUMSTANCE that covers a

multitude of adverbial meanings (manner, degree, condition, background,
intention, text-relation, speaker-relation).
I) Role suppression. There is in English a tendency for fusion between the
predicator and one or more of the following constituents (cf. section 4.3 on
complex predicators). In our discussion of the S P O C pattern we thus noted
that there are varying degrees of attachment between the complement and the
predicator: sometimes, as in Tyson knocked Bruno unconscious, there is
interdependence, sometimes, as in Roger painted the wall blue, the
complement seems much freer. When a constituent is very closely fused with
the predicator, both constituents may lose some of their semantic
independence. In turn, this may affect the assignment of semantic roles, not
only to the fused constituent itself, but also to other constituents in the
sentence. Consider first an example like the following:
(42) I caught sight of her.
Here it would be counter-intuitive to assign a separate participant role to
sight: Rather, it would be reasonable to analyse caught sight o/as a complex
predicator (= 'sighted') and simply analyse her as AFFECTED object.
The S P O and S P O O patterns are especially prone to fusion between
predicator and direct object when the predicator is realized by semantically
general verbs like GIVE, TAKE, DO, HAVE, MAKE. The effect on the S P O
pattern is to make it 'semantically intransitive':
(43) They had an argument, (cf. They argued)
(44) She made a complaint, (cf. She complained)
The effect on the S P O O pattern is to turn it into a 'semantically mono-
transitive' construction. GIVE is especially frequent in this pattern:
(45) She gave her daughter a smack/nudge/bath/kiss, (cf. She smacked/nudged/
bathed/kissed her daughter)
(46) I paid my mother a visit, (cf. I visited my mother)
As a result, the semantic function of the indirect object is in such cases not
BENEFICIARY but AFFECTED.
Finally, mention should be made of cases where the predicator is realized
by a lexical verb embodying an INSTRUMENT (cf. Radden 1989: 44If):
(47) The assassin knifed the President.
(48) I spoon-fed her daughter.
(49) She winked humorously.
(50) I mouthed the insult behind her back.
Voice 205
Many verbs imply the INSTRUMENT with which the situation referred to is
usually brought about: e.g. GRAB, SMILE, WADE, STAB, SWEEP, etc.
7.4. Voice
7.4.1. Introduction
In English, a sentence in the passive voice is a sentence in which the
realization of the predicator contains a form of the auxiliary BE followed by
the -ed participle form of the main verb and in which the subject form
prototypically performs the participant role DONE-TO (see section 7.3):
(1) The county prosecutor was finally prodded into action.
The terms 'active' and 'passive' are based on the semantic function performed
by the subject form in sentences describing dynamic situations. While this
form denotes the active participant (DOER) in an active sentence like Our
boss will kill me, it denotes the passive participant (DONE-TO) in a passive
sentence like Our boss was killed in a plane crash.
In passive sentences the DOER may be specified by a prepositional by-
group:
(2) No public explanation or apology was made by the Hazelton police.
If this sentence is compared with the active sentence The Hazelton police
made no public explanation or apology, we see that voice is to do with the
way in which the forms referring to the participants involved in the situation
described are syntactically related to the verb. In this sense voice is a
category of the entire clause, and our account of it is therefore given in the
present chapter on the simple sentence. But as voice also involves two
different realizations of the predicator (e.g. made vs. was made), it is
simultaneously a category of the verb group, and in this respect it resembles
tense (e.g. makes vs. made).
Each of the four transitive patterns found in active declarative sentences
(cf. section 3.2.4) has a passive counterpart. In the case of S P O, S P O Co
and S P O A, the passive counterparts are S P (A), S P Cs (A) and S P A (A):
(3) She might have been saved (by her fellow passengers).
(4) He was ruled mentally unfit (by the judge).
(5) They were placed on their backs (by the robbers).
A passive sentence containing a DOER by-group is roughly synonymous
with the corresponding active sentence. For example, He was ruled mentally
unfit by the judge has the same prepositional meaning as The judge ruled him
mentally unfit.
The passive counterpart of S P Oi Od is S P O (A):

(6) The butler was given a reward (by the police).
As can be seen, it is the noun group realizing the indirect object of the
corresponding active sentence which functions as subject. For some speakers
of English it is also possible to select as subject form in a passive S P O (A)
sentence a noun group which in the active sentence realizes the direct object:
(7) A reward was given the butler (by the police).
This type of construction, however, is much rarer than A reward was given to
the butler (by the police).
A large majority of passive sentences do not contain a fty-group, and this
type of adverbial is practically never obligatory. Examples like Elizabeth I
was preceded by Henry VIII and Henry VIII was followed by Elizabeth I are
thus exceptional in that their final preposition group cannot be omitted.
Passive sentences may also contain complex predicators:
(8) Her personal effects had been parcelled out to her sons after her death.
(9) The restaurant was surprisingly crowded but they were waited on quickly.
(10) This must be taken care of immediately.
In closing this section we should draw attention to passive sentences with
complex predicators like the following (cf. van Ek & Robat 1984: 246f):
(11) Great fun was made of his remark.
This sentence may be compared with His remark was made fun of, which
behaves like (10) in that its subject is realized by the form which in the
corresponding active sentence functions as object of the complex predicator:
People [made fun of] [his remark]. If the noun of a complex predicator has a
pre-head dependent (a quantifier or a modifier), as in People [made great fun
of} [his remark], however, it is this noun which together with its pre-head
dependent realizes the subject of the passive sentence, as in (11). Other
examples illustrating this difference are You have been taken advantage of
and Little advantage has been taken of this opportunity.
7.4.2. Functions of the passive

The passive provides a systematic means of choosing another participant
than DOER as starting-point for the message without departing from subject-
first constituent order (see Johansson & Lysväg 1986: 99 and section 7.3
above). It should be recalled that through fronting of objects, complements
and adverbials we also select another participant than DOER as starting-
point, as illustrated by That story I will tell you another time, Chris Waddle
Voice 207
his name is and Here comes the bus respectively. But in such cases of
fronting the preferred subject-first order is not retained.
In other words, the use of the passive vs. active voice in English is largely
determined by the way the speaker wishes to organize his message, i.e. is to
do with information structure. In a passive sentence like Tottenham were
beaten by Arsenal, where the participant AFFECTED DONE-TO is placed
initially, it is this participant (Tottenham) and not AGENT DOER (Arsenal)
which constitutes the speaker's communicative point of departure.
A main reason for choosing the passive voice is thus thematization (or
topicalization, see sections 3.2.2 and 5.2.3): a form performing the
participant role DONE-TO is placed initially - where it realizes the subject -
in order to make it the theme (or topic) of the message. The passive sentence
Tottenham were beaten by Arsenal differs from the corresponding active
sentence in that it is a statement 'about' Tottenham, not Arsenal. While the
two sentences have the same prepositional meaning and the same
illocutionary force, they differ with respect to thematic meaning. Thematic
meaning has been defined as "what is communicated by the way in which a
speaker or writer organizes the message, in terms of ordering, focus and
emphasis" (Leech 1981: 19) and shown primarily to involve choice between
alternative syntactic constructions.
A second main reason for selecting the passive voice is to avoid
mentioning the DOER participating in the situation described. This is
illustrated by the following example, which simultaneously illustrates
thematization in that it is a text 'about' a specific person:
(1) Evander Jones is a war veteran, was awarded a Purple Heart, honorably
discharged at the end of the war. Nine years ago he was convicted of first-
degree murder in a drugstore holdup and sentenced to death.
By means of the passive the speaker is at liberty not to provide information
which has to be provided in active sentences. As DOER is often unknown,
irrelevant, unimportant or can be inferred from the linguistic or situational
context, this makes the passive an extremely useful construction. In (1) it
saves the sender the trouble of specifying who awarded Evander Jones a
Purple Heart, discharged him, convicted him and sentenced him.
As pointed out in section 5.2.3, the organization of a message often reflects
a division between given and new information, and the choice of the passive
may also be due to a wish to proceed from given information:
(2) The front of the station waggon was crushed like an accordion by a big
boulder.
This principle obviously cannot apply to sentences containing new
information only, as illustrated by Residents have drawn up and signed a
petition and A petition has been drawn up and signed by residents in which
there is no given material to proceed from.
The passive may also be selected to obtain end-focus or end-weight
(3) The last cup final was won by Newcastle.
(4) The hearing was undercut by Edward's refusal to testify and Thiel's obvious
reluctance to provide jurors with information.
In narrative, the passive may be selected in order to retain the same subject
in successive clauses. In this way the presentation of new before given
information is avoided. The selection of the passive here may also be
connected with a wish to avoid mentioning the DOER or with the principle
of end-weight:
(5) He demanded to speak to her but was refused.
(6) The case received a good deal of publicity locally and was taken up
immediately by the state branch of the ACLU.
Finally it should be mentioned that passives may be motivated by a wish on
the part of the speaker to avoid self-reference and come across impersonally:
(7) Enough has been said above about the implications of the Faculty's
announcement for the future of our Ph.D. programme.
In scientific English, where it is typically what happens which is of interest
rather than who makes it happen, passive constructions (without DOER by-
groups) tend to be more frequent than in other registers of English. This
usage can be illustrated by the following example (quoted from Collins
Cobuild English Grammar 1990: 404):
(8) Food is put in jars, the jars and their contents are heated to a temperature
which is maintained long enough to ensure that all bacteria, moulds and
viruses are destroyed.
7.4.3. Extended use of the passive

As we have seen, the voice category involves a regular correspondence
between a transitive construction with a DOER subject, an active predicator
and a DONE-TO object and an intransitive construction with a DONE-TO
subject, a passive predicator and an optional DOER 6y-group:
(1) John caught the rat.
(2) The rat was caught (by John).
The members of the voice category are thus typically alternative syntactic
expressions of dynamic situations. In this section we examine two extensions
of this basic model.
Voice 209
A) Sentences describing stative situations. The most important extension

of the passive in English concerns sentences describing stative situations.
Active transitive sentences of this kind are normally unmatched by passive
sentences. This can be illustrated by the following examples:
(3) Ildiko has a red Mercedes Benz, (extensive relation)
(4) She hates all the fuss, (attitude)
(5) I felt her damp hair against my skin, (perception)
(6) Roger collects stamps, (habit)
As pointed out in section 7.2.3, stative situations are almost always relations
of some sort, i.e. the general participant roles involved are not DOER and
DONE-TO but SPECIFIED and SPECIFIER. Now, in a number of cases the
passive may be extended to sentences describing stative situations and in
which the subject form is AFFECTED SPECIFIER. The situational subtypes
involved are attitudes and perceptions:
(7) She is feared and resented for her outspokenness.
(8) This rumour is widely believed.
(9) She's been seen at night in neighbourhoods unsafe for solitary women.
(10) A shot was heard in the dark.
A possible reason why such passive sentences are found is that the situations
they describe are not entirely stative. For example, the situations described
by (7) through (10) require a certain input of energy (see section 7.2.1), like
those described by active sentences such as People in the neighbourhood
believe this rumour (attitude) and Several witnesses have seen her at that
hour (perception). But even where attitudes or perceptions are involved, the
passive does not have free play, as illustrated by (4) and (5). This particularly
goes for sentences describing attitudes. For example, John believes the
rumour - in which the SPECIFIED is identified very precisely - can hardly
be said to have a natural passive parallel. In sentences describing perceptions
- and in which the SPECIFIED is identified precisely - the passive is
sometimes acceptable and sometimes problematic or unacceptable. While
e.g. Sally was seen by John can hardly be ruled out as an alternative to John
saw Sally, a sentence like John smelled the flowers appears to be unmatched
by any passive sentence.
The passive is particularly frequent in stative examples like the following:
(11) They were assumed to be dead.
(12) Andy was considered unfit for fight.
(13) Clyde is known by everybody as a good swimmer.
In these examples, the constituent in subject position receives a double

participant role analysis. In (11), which is a 'nominative with infinitive1 (cf.
section 5.6.1), They is not simply SPECIFIER AFFECTED in relation to the
attitudinal state of 'assuming' but also, at a secondary level, SPECIFIED
HOLDER of the ATTRIBUTE dead (cf. section 7.3). And in (12) and (13),
Andy and Clyde receive a similar analysis.
Before proceeding to another passive extension, we should draw attention
to sentences like Her dress touched the ground, which has no passive
parallel, and The intruders haven't touched the merchandise, which is
matched by The merchandise hasn't been touched by the intruders (see
Huddleston 1984: 439). The first of these describes a stative situation of the
subtype extensive relations, so here the passive is regularly ruled out. The
second, on the other hand, describes a dynamic situation involving the
participant roles DOER and DONE-TO, so here the passive is regularly
permitted. What the examples show is that the same lexeme (TOUCH) may
be used for the description of different situations. Other verbs behaving in
this way are HAVE and POSSESS. Normally they are ruled out in passive
constructions, but when used in sentences describing dynamic situations they
are permitted, as illustrated by Cheese can be had in the store across the
street and He is possessed by demons.
B) Sentences with 'stranded* prepositions. Another extended use of the
passive can be illustrated by examples like the following:
(14) The bed has been slept in.
(15) Their house hasn't been lived in for a long time.
It seems most reasonable to assume that these examples do not contain
prepositional predicators, i.e. predicators which in active sentences are
followed by objects. There is no clear semantic fusion between verb and
preposition, and the preposition is not the only one associated with the verb
(cf. section 4.3.2). What we have here therefore appear to be passive
sentences with a stranded preposition matched by intransitive active
sentences. For example, the nearest active parallel of (14) is Someone has
slept in the bed, which can only be classified as an S P A construction. In
such cases, the subject of the passive sentence corresponds not to an object
but to the prepositional complement in the adverbial of the active sentence.
7.4.4. Voice restrictions

So far we have established that transitive sentences in the active voice are
often matched by intransitive sentences in the passive voice, and vice versa
(e.g. The cat chased the mouse : The mouse was chased by the cat). Because
Voice 211
of this highly regular pattern, it has often been assumed that there is a fixed,
automatic correspondence between active and passive sentences. However,
this view of the voice category is too simplistic. A number of important
restrictions apply to both active and passive constructions.
A) Passives without corresponding actives. In the large majority of cases
where a passive sentence contains no DOER by-group, there is no specific
corresponding active sentence. In The county prosecutor was finally prodded
into action, for example, we do not know what form the subject of the
corresponding active sentence is realized by (somebody? something? people
in the community?). The subject form may be recoverable from the
situational context, but in very many instances it cannot be recovered from
the linguistic context. In some cases, furthermore, it is virtually impossible to
supply a corresponding active construction with a subject form at all: e.g.
Clyde was known locally as a very good swimmer.
Passives without corresponding actives are also found where a small group
of specific verbs are involved. For example, sentences with BE born, like
Shakespeare was born in 1564, have no active parallels. When followed by
to-infinitive constructions, similarly, SAY and SEE are passive only: while
we can say e.g. She is said to be very rich and She was seen to leave the
house (with discontinuously realized subject clauses, see section 5.6.1), we
cannot say *They say her to be very rich and *They saw her to leave the
house (though the latter becomes acceptable if the infinitive marker is
removed). What we find here are nominative with infinitive passive
sentences unmatched by accusative with infinitive active sentences (cf.
section 5.6.1). Attention should also be drawn to nominative with infinitive
constructions with BE rumoured and BE reputed like She is rumoured to
have shot him and He is reputed to be very rich. Again there are no active
counterparts and a by-group cannot be inserted. Finally it should be
mentioned that a number of verbs are largely though not wholly restricted to
passive sentences. This goes for ACCLAIM, SCHEDULE and phrasal verbs
like BOWL over (in the sense of 'surprise') and TAKE aback (e.g. Our next
meeting is scheduled for Monday and / was completely taken aback by her
behaviour).
Consider next an example of the following type:
(1) Princeton has been visited by Einstein.
Here there is no corresponding active sentence either, for *Einstein has
visited Princeton is normally unacceptable if pronounced with unmarked
prosody, i.e. with nuclear stress on Princeton. The reason for this is that it
would be a sentence 'about' Einstein, and in English the present perfect
cannot normally be used if the subject form realizing the topic (or theme)
refers to a deceased person. If Einstein is uttered with nuclear stress and

Princeton with weak stress, the active sentence is acceptable, for by means of
such marked prosody the speaker indicates that the former word constitutes
the comment and the latter (part of) the topic.
B) Actives without corresponding passives. In the first place, restrictions
on the passive may be of a formal nature. If an active sentence combines the
progressive marker BE + -ing with a modal verb or/and the perfect marker
HAVE + -ed - as in The cat must be chasing the mouse, The cat has been
chasing the mouse and The cat must have been chasing the mouse - a
corresponding passive sentence would contain two nonfinite forms of BE.
Such a combination is decidedly rare: ?The mouse must be being chased by
the cat, ?The mouse has been being chased by the cat, ?The mouse must have
been being chased by the cat. In those cases where the perfect occurs, it may
be added, speakers nearly always match a progressive active sentence with a
nonprogressive passive sentence, that is, they select e.g. The mouse has been
chased by the cat and The mouse must have been chased by the cat instead of
The mouse has been being chased by the cat and The mouse must have been
being chased by the cat. Very occasionally, however, one does come across
sentences with two nonfinite forms of BE and in some cases even sentences
with four auxiliaries (see section 3.2.1 on the extreme rareness of this):
(2) Not all instances of a given strategy may be being used to do politeness.
(3) ... another £98m was paid direct by absent parents to their former parents.
Much ofthat could well also have been being paid before April last year, Mr
Field argued.
Though passive sentences with two nonfinite forms of BE thus cannot be
ruled out, they are so infrequent that it is reasonable to speak of a restriction
on the passive voice here. And particularly perfect progressive forms in the
passive are very rare indeed.
With respect to sentence type, it should be mentioned that the passive
occurs in declarative sentences, interrogative sentences (e.g. Has a public
apology been made?) and exclamatory sentences (e.g. How elegantly I'm
received!) but that it is heavily restricted in imperative sentences, which are
typically exhortations to the addressee to act. Only in fixed expressions like
Please be seated and in those cases where the hearer is ordered or advised to
avoid becoming the target of an action, as in Don't be taken in by that
scoundrel, do we find imperative sentences in the passive voice.
Constraints on the passive may also be due to the nature of the object of an
active sentence (see e.g. Johansson & Lysväg (1986: 103ff), who give a
detailed account of such constraints). If the object is cognate, i.e. if the noun
realizing its head is derived from the verb preceding it as in live a good life
Voice 213
and sleep the sleep of the just, the passive voice is normally ruled out. Nor do
fused Ρ Ο constructions like take a bath and have a smoke have passive
counterparts. Furthermore, if there is coreference between the subject and
object forms, the passive voice is usually also excluded. For example,
sentences with reflexive or reciprocal object pronouns like He shot himself
and We hate each other have no passive counterparts. As reference is here
made to the same person(s), there is no thematic reason why an alternative
arrangement should be preferred.
Restrictions on the passive may also be due to the fact that the object of a
transitive (complex) sentence is realized by a clause. As illustrated by e.g.
That he is clever is known by all his friends and colleagues the passive is not
impossible here. But most sentences with objects realized by finite clauses
have no natural passive counterparts. Only by means of extraposition -
which secures end-weight - is it possible to match an example like The
public believe that there will be an election with a passive: // is believed by
the public that there will be an election (see section 5.2.3 on extraposition
and end-weight). In those cases where the object of an active sentence is
realized by a nonflnite clause, the passive is heavily restricted too. For
example, sentences like / would like to visit you, I don't like interrupting you
and / would like you to visit me are unmatched by passive sentences.
In those cases where the predicator of an active sentence is realized by a
reciprocal verb, such as MARRY, EQUAL, MEET and RESEMBLE, there is
usually no corresponding passive sentence (see Huddleston 1984: 440). Here
there is little need for the passive, for if the speaker wishes to thematize the
entity referred to by the object form, all she needs to do is to interchange this
form with the subject form. Instead of saying John married/met/resembles
Mary she can simply say Mary married/met/resembles John. This thematic
strategy is not available in a case like John resembles a poodle where the
object form refers to a category. Note also examples like Tyson has only been
equalled by one fighter and Mary was met by John in the airport, which
illustrate that these verbs are not completely ruled out in passive sentences.
Attention should also be drawn to sentences which contain both a negative
element and one or more so-called nonassertive forms (i.e. forms typically
restricted to negative and interrogative sentences like ever, any, either and
yet). For example, sentences like The committee hasn't accomplished
anything and She hasn't ever kissed anyone are unmatched by passive
sentences composed of the same words, as shown by the ungrammaticality of
*Anything hasn't been accomplished by the committee and * Anyone hasn't
ever been kissed by her. Here the nearest passive constructions are Nothing
has been accomplished by the committee and No one has ever been kissed by
her. We return to the ordering of negative elements and nonassertive forms

in section 7.5 below.
Consider finally examples like the following:
(4a) Many arrows didn't hit the target.
(5a) We won't do the job.
Here it is easy enough to construct the following passive sentences:
(4b) The target wasn't hit by many arrows.
(5b) The job won't be done by us.
It will be noticed, however, that the propositional meaning of the passive
sentences is different for each pair from that of the active sentences. In (4a)
and (4b) this is due to the semantic effect of negation (cf. section 7.5 below):
by means of (4a) - but not (4b) - the speaker may describe a situation where
the target was in fact hit by many arrows, for example 50 out of 100. Though
many missed, many others didn't. In (5a) and (5b) the difference in meaning
is due to the modal verb. In (5a) won't means 'are unwilling to* or 'refuse to',
but in (5b) it will in most situations be understood to signify a prediction of a
future event. In other words, (un)willingness is not involved here.
7.4.5. Nonfinite passives, GET-passives and notional 'passives'

In complex sentences - which we have chosen to touch on occasionally to
confine our discussion of the passive to one chapter - the predicator of a
subordinate passive clause is often nonfinite (see section 3.2.1):
(1) What the boy had done now - or failed to do - Lee hoped not to be told.
(2) A few days after being released from jail he disappeared.
(3) Nowhere in this place could she move among people who knew her without
being critically observed, assessed, commented upon.
As appears from (1) through (3), nonfinite passive verb groups are either
infinitive or -ing participle constructions and, in addition to to be + V + -ed
and being + V + -ed, we also find the perfect forms to have been + V + -ed
and having been + V + -ed.
Vacillation between passive and active occurs after the main verbs BE,
NEED, DESERVE and WANT, as in There is no time to lose/to be lost, It
needs doing/to be done, She deserves punishing/to be punished and Your
aspidistra wants watering/to be watered. Note in this connection examples
with REQUIRE and BEAR like My car requires servicing and This does not
bear repeating; although the active form here also has passive meaning, only
the active voice is possible.
Voice 215
In section 7.4.1 we stated that in passives the predicator is realized by a form

of the auxiliary BE followed by an -ed participle form of the main verb. It
should be added, though, that the predicator of a subordinate passive clause
is sometimes realized by an -ed participle form exclusively:
(4) Lydia knew herself -watched.
(5) We saw Denmark beaten by Spain.
This type of construction - in which being could be inserted at the beginning
of the nonfinite predicator - is restricted to those cases where the finite
predicator is realized by a small group of verbs, such as KNOW, SEE,
WATCH, HEAR and WANT (see Huddleston 1984: 444). Attention should
also be drawn to passives without BE like the following:
(6) Considered unqualified for the job, she was asked to leave on the spot.
Note that being could be inserted here as well. Participle -ed clauses are not
invariably passive, it should be added. For example, the subclause of a
sentence like Escaped from prison, the convict immediately contacted his old
partners in crime is clearly active (Escaped = Having escaped).
Another case in which BE is absent from a passive sentence occurs where
it is a form of GET which precedes the -ed participle:
(7) Dreyfus got hit on the head with a rake.
(8) Denmark got beaten by Spain.
A GET-passive typically serves to express that the referent of the subject
form passes from one state to another, as in The thief got caught in the end.
This appears to be due to the meaning of GET, which has been described as
"arrive at a resultant state" (Palmer 1987: 89). Consequently, there are many
instances where GET cannot replace BE While e.g. Denmark were beaten by
Spain and Denmark got beaten by Spain are equally acceptable, the finite
verb cannot be replaced by got in examples like A public hearing was held
and Through the spring they were often seen together in Yewville.
GET-passives differ from BE-passives in that an element of initiative or
responsibility is often ascribed to the referent of the subject form, as
illustrated by Malcolm got promoted/arrested (see Huddleston 1984: 445).
This obviously only applies to those cases where such a referent is human,
but GET-passives like The jug got broken in which the referent of the subject
form is non-human are relatively rare. Note in this connection the possibility
of using GET-passives in the imperative: Get lost! I Get (yourself) invited to
the meeting (cf. Halliday 1994: 76).
In some contexts, the GET-passive has the advantage over the BE-passive
that it is unambiguous. In examples like The jug got broken and They got
married it can only be dynamic situations which are described. On the other
hand, The jug was broken and They were married can refer either to a
dynamic situation or a state. In the former case they are passive S P
sentences but in the latter they are active S P C sentences in which the
complement is realized as an adjectival participle (see section 7.4.6 below).
In BrE, GET-passives are relatively infrequent and restricted to colloquial
style. In AmE, according to Granger 1983: 235, they are becoming very
common among young people, particularly males and blacks.
In section 7.4.2 we pointed out that the passive provides a systematic
means of choosing another participant than DOER as starting-point for the
message (thematization) and that it permits the speaker to avoid mentioning
the DOER participating in the situation described. Instead of saying
Somebody opened the door we can thus say The door was opened. A fairly
large class of verbs also appear in intransitive active sentences: The door
opened. As this is semantically similar to the (syntactic) passive, it is
sometimes termed the notional passive. If The door opened is compared with
Somebody opened the door, we see that conversion from transitive to
intransitive is accompanied by thematization of the door and that somebody
is not only gone but also incapable of appearing in a fey-preposition group.
As pointed out in section 7.3.4 [A], however, The door opened differs from
The door was opened in that the participant chosen as starting-point for the
message is not DONE-TO but DOER. By means of the intransitive
construction the speaker presents the situation as if the subject is DOER
while at the same time being aware that it cannot do this of its own accord.
What is similar to the two constructions, though, is that there is no
prepositional fey-group and that the specific participant role performed by the
subject is AFFECTED. Although the semantics of the two constructions is
not the same, we are thus likely to understand them in terms of similar
situations.
The verbs found in notionally 'passive' sentences will here be called middle
verbs. Such verbs can take the same form as their object or subject, in
transitive and intransitive sentences respectively. As the intransitive
sentences involved are both (syntactically) active and passive, middle verbs
involve trios of constructions, as illustrated by Somebody opened the door,
The door was opened and The door opened (see section 7.3.4 [A]).
The intransitive use of middle verbs in examples like The door won't lock,
Sugar dissolves in water, The shop has closed and This stanza doesn't scan is
lexically restricted, but the class of middle verbs in English is by no means a
small one (see Collins Cobuild English Grammar 1990: 155ff). In addition to
verbs like OPEN, CLOSE, LOCK, BREAK, CRACK, SHATTER and WIDEN,
there are verbs relating to cooking like BAKE, BOIL, COOK, FRY and
ROAST (e.g. The eggs are frying) and verbs which combine with a few
Voice 217
specific subject forms only, like FIRE, SHOW and SOUND (e.g. The pistol
fired, Her fatigue showed and The bugle sounded). Some middle verbs are
nearly always accompanied by adverbials (adjuncts), usually of manner:
(9) Julia Roberts photographs well. (= is photogenic)
(10) This loaf doesn't cut easily.
Note finally the use in BrE of BE drowned and BE burned:
(11) Shelley was drowned.
(12) The house was burnt down.
These are more or less synonymous with the now more frequent active
middle-verb construction Shelley drowned and The house burnt down, i.e.
without a sense of there being a suppressed agent: in both the active and the
passive construction the subject is AFFECTED (by water or by fire). But
there is also a truly passive interpretation possible in formally passive
constructions with an overt or implied AGENT: The kittens were drowned
(by the two boys) / The palace was burnt down (by the rebels) corresponding
to The two boys drowned the kittens / The rebels burnt the palace down.
7.4.6. Passives vs. adjectival non-passives

In English it is often the case that a sentence containing a form of BE
followed by an -ed participle is not passive but active:
(1) I am not accustomed to being interrupted.
(2) All these people are educated and very reliable.
(3) My heart was so swollen with feeling I could not reply.
Here BE is not an auxiliary but a main verb, and the -ed participle form does
not realize the main verb but (part of) the subject complement. The situations
described are stative, and the participles are not verb forms but adjectives
derived by lexical-morphological conversion of a verbal -ed form.
As pointed out in section 7.4.5, sentences like The jug was broken and
They were married are ambiguous in that they can refer either to a dynamic
situation or to a state. In the former case they are passive S P sentences and
in the latter case active S P C sentences. In sentences like The ship was sunk
and His cheeks were sunken (cf. Schibsbye 1970: 18) the difference between
passive and active is signalled formally. Like sunken, the word shaven is
exclusively used as an adjective, but here the other form shaved need not be
passive when preceded by a form of BE. Out of context a sentence like He
was shaved is ambiguous between a dynamic and a stative reading.
Adjectival -ed forms differ from passive participles in a number of ways
(see e.g. Palmer 1987: 85f and van Ek & Robat 1984: 252):
(i) They may be used after verbs other than BE, such as SEEM and BECOME.
(ii) They accept intensifiers like VERY, RATHER, etc., the markers of
comparison MORE and MOST and negative prefixes. (Non-gradable -ed
forms like married, however, cannot normally be intensified or compared.)
(iii) They may be coordinated with a true adjective.
(iv) Insertion of ALREADY in sentences in which they occur does not
necessitate the use of the perfect.
The sentence pattern encountered in an example like The case is complicated
is thus S P C, not S P (see Palmer 1987: 86):
(i) The case seems/is becoming complicated,
(ii) The case is quite complicated.
The case is more complicated than we expected.
The case is uncomplicated,
(iii) The case is awkward and complicated,
(iv) The case is already (quite) complicated.
On the other hand, a sentence like Three of the passengers were saved is
passive, for here the -ed form has none of the properties listed in (i) through
(iv), and the same goes for e.g. Plans for an amnesty were complicated by
renewed terrorism. Note that insertion of ALREADY in these two examples
would necessitate a change of were to had been.
While the distinction between passives and adjectival non-passives is
basically clear, it should not be overlooked that borderline cases do exist and
that a grey zone has to be reckoned with in this grammatical area as in so
many others. In The students were amused/annoyed/embarrassed by my dirty
jokes, for example, the existence of the corresponding active construction My
dirty jokes amused/annoyed/embarrassed the students points towards a
passive analysis. On the other hand, the possibility of replacing were by
seemed and of premodifying the participle with an intensifying adverb, a
comparative marker or a negative prefix (e.g. quite amused/more amused
than I'd expected/unamused) seem to indicate that the -ed forms are
adjectival non-passives instead. The presence of a by-group, it will be noted,
is thus not a sure sign that the preceding -ed participle is passive (see e.g.
Quirk et al. 1985: 169). This appears very clearly from an example like The
students were unamused by my dirty jokes, which can only be analysed as an
S P C sentence whose complement is realized by an adjective group. Note
also that the head of the postmodifying preposition group could be realized
by at instead of by with little or no change of meaning. Such prepositional
Polarity 219
variation can be further illustrated by worried by/about, satisfied by/with and

disappointed by/in/at/about /with.
7.5. Polarity
7.5.1. Introduction
One of the optional adverbials that may expand the basic sentence patterns
(cf. sections 3.2.4 and 3.2.9) is the negative adverb NOT:
(1) Software is not like other intellectual property.
This example is a negative sentence. The distinction between positive and
negative is one of polarity. Semantically, a negative sentence differs from a
corresponding positive sentence in expressing that one or more of the
conditions which must be satisfied in order for this positive sentence to be
used appropriately are not satisfied. For a positive sentence like Sanctions
challenge vital interests to be used appropriately, a number of conditions
must be satisfied (see Huddleston 1984: 432). These conditions represent
different layers of meaning: (i) Something challenges something, (ii)
sanctions challenge something, (iii) something challenges interests, (iv)
sanctions challenge interests, (v) something challenges vital interests, (vi)
sanctions challenge vital interests. When the corresponding negative
sentence Sanctions don't challenge vital interests is used, the speaker
expresses that one or more of these conditions is not satisfied. If it is (ii), (iv)
and (vi) which are not satisfied, this can be indicated by pronouncing the
word sanctions with contrastive stress:
(2) 'Sanctions don't challenge vital interests.
If it is conditions (v) and (vi), on the other hand, this can be shown by
placing the nuclear stress on the adjective vital:
(3) Sanctions don't challenge 'vital interests.
What is indicated prosodically in (2) and (3) is focus of negation, i.e. the
place where the contrast of meaning expressed by the negation is located (the
subpart of the sentence expressing the condition that is not satisfied). In (2),
for example, it is only the referent of the subject noun which is at issue, not
the truth of the rest of the sentence. This focus can be made explicit by using
a so-called cleft sentence instead (cf. section 8.4):
(4) It isn't sanctions that challenge vital interests.
Such a construction can naturally also be used in spoken English. In that case
sanctions will still be pronounced with nuclear stress, so here the focus of
negation is doubly marked.
Negative sentences can also be pronounced 'neutrally', i.e. with the nuclear
stress on the last content word - e.g. interests in (4) - and without stress
reduction in the preceding part of the sentence. In that case it is not specified
exactly what condition or conditions are not satisfied, and it therefore
becomes a matter for the hearer to work this out on the basis of the context.
Note finally that a negative sentence may be used as a denial, i.e. as a
refusal to accept as a fact a previous statement. If speaker A says e.g. that
Sanctions challenge vital interests, speaker B may deny this statement by a
negative sentence in which don't is pronounced with nuclear stress:
(5) Sanctions 'don't challenge vital interests.
The difference between denials and Ordinary1 negative statements sometimes
sheds light on difficult examples. For instance, a sentence like They don't
owe me nothing can in Standard English only be used as a denial of They owe
me nothing, i.e. roughly in the sense They owe me something'. Outside
Standard English, such an example can also be used as an ordinary negative
statement, i.e. with the same meaning as They don't owe me anything.
7.5.2. Standard negation

In its 'standard' version, negation is signalled by placing NOT after the
operator, as illustrated by The cycle of death will not continue. If there is no
operator in the corresponding positive sentence, as in Sanctions challenge
vital interests, the dummy operator DO is inserted before NOT in the negative
sentence. In informal English the negative adverb is typically contracted
with the preceding operator. In that case it is pronounced /nt/ and written n't:
(1) They shouldn't blame us.
(2) I didn't know much about Rwanda.
When contracted with CAN, it should be added, NOT can also be written as
not and pronounced - without stress - as /nOt/:
(3) Without autonomy, we can't/cannot be a truly free and democratic society.
The only auxiliary which NOT may not be freely contracted with is MAY.
The contracted form mayn't does not occur in AmE, and in BrE it is
extremely rare.
In those cases where the predicator is realized by a form of the full verb BE
or by a complex verb group in which the operator is a form of BE, HAVE or
WILL, the negative adverb may thus be contracted with the preceding verb
form like the modal auxiliaries and DO (compare examples (1) through (3)):
(4) It isn't true.
(5) She hasn't resigned yet.
Polarity 221
(6) He won't be missed much.

Here, however, there is another alternative: the finite verb may be contracted
with the preceding subject realization instead (verb contraction instead of
NOT-contraction). When this happens, NOT is obligatorily stressed:
(7) It's'not'true.
(8) She's 'not re' signed yet.
(9) He'll "not be 'missed much.
This type of contraction is particularly common if the subject is realized by a
pronoun, but in the shape of 's BE and HAVE are often contracted with a sub-
ject nominal too, and informally WILL sometimes behaves that way as well:
(10) The answer's not especially complicated.
(11) Peter's not resigned yet.
(12) Pete'll probably not accept the deal.
While DO and the modal auxiliaries (except WILL) permit only NOT-
contraction, the verb form AM permits only verb contraction, as in I'm not
sorry and I'm not studying. In non-standard English, particularly non-
standard AmE, the contracted form ain 't is used for am, are, is or has/have +
NOT, for example in I/she/you/they ain't sorry and I/she/you/they ain't
finished yet. Finally it should be mentioned that the contracted form aren't
may be used in BrE not only with 2nd person singular or with plural subjects
as in You aren't the only survivor and They aren't here but also with 1st
person singular subjects. This use is restricted to negative questions, though,
and is particularly common in tag questions: I'm your wife, aren't I?
7.5.3. Domain of negation

In examples with standard negation, the domain of negation is global in the
sense that the unit affected semantically by NOT is the entire clause. Global
negation as in You can't come tomorrow is analysed in the following way:
STA
tomorrow
can come
In order to separate global from non-global negation a number of criteria can

be used (see Klima 1964 and Payne 1985: 198). It is a characteristic of a
sentence with global negation that it permits:
(i) a positive rather than a negative tag question;
(ii) a coordinated clause beginning with (and) NEITHER rather than (and) SO
or ending in EITHER rather than TOO;
(iii) a construction beginning with NOT EVEN.
According to these criteria a sentence like You can't come tomorrow with
contraction has global negation:
(1) You can't come tomorrow, can you?
(2) You can't come tomorrow and neither can your wife.
(3) You can't come tomorrow and your wife can't come either.
(4) You can't come tomorrow, not even in tails.
If there is no contraction and NOT is pronounced with stress, the syntactic
structure of the sentence may be the same (i.e. S P- A -P A), and apart from
the degree of emphasis involved so may its meaning. However, You can not
come tomorrow differs from You can't come tomorrow in being ambiguous,
for it can be used either to express that the person addressed is "not permitted
to come' or that he is 'permitted not to come'. The latter meaning can be made
clear by the addition of adverbials, as in You can always not come tomorrow
if you like. Here the negation is not global but local. In sentence analysis this
can be captured by using the stack convention (cf. section 4.1):
STA
cl
S £ A
pro g adv
You PEP H tomorrow

V X
can PEP H
adv v
I I
not come
That negation is non-global here appears from application of the criteria: the
clause can be followed by a negative rather than a positive tag question, or
by a coordinated clause beginning with (and) SO rather than (and) NEITHER
or ending in TOO rather than EITHER, and it cannot be followed by a
construction beginning with NOT EVEN:
Polarity 223
(5) You can not come tomorrow, can't you?

(6) You can not come tomorrow and so can your wife.
(7) You can not come tomorrow and your wife (can not come) too.
Global and local negation are capable of cooccurring, for example in You
can't not come tomorrow (= You are not at liberty to stay away tomorrow).
While the combination of global and local negation is stylistically marked
here, it is natural in those cases where local negation is realized by a negative
prefix, i.e. where negation operates within the boundaries of the word:
(8) Gates doesn't address these matters irreverently.
(9) He isn't unhappy about his work.
Global negation is not an exclusive property of standard negation. For
example, we find it not only in a sentence like The earthquake didn't cause
any casualties but also in a sentence without standard negation like The
earthquake caused no casualties. That this is so appears from the fact that
the second of these examples, like the first, selects a positive tag question,
combines with NEITHER and EITHER instead of SO and TOO and permits a
following NOT EVEN construction:
(10) The earthquake caused no casualties, did it?
(11) The earthquake caused no casualties and neither did the ensuing landslide.
(12) The earthquake caused no casualties and the ensuing landslide caused no
casualties either.
(13) The earthquake caused no casualties, not even around the epicentre.
While The earthquake caused no casualties thus resembles The earthquake
didn 't cause any casualties in having global negation, it differs from it with
respect to syntactic field, a subject which we now turn to.
7.5.4. Syntactic field of negation

While the domain of negation concerns the overall polarity of clauses,
syntactic field concerns the syntactic material actually negated. In an
example with global negation like The earthquake caused no casualties, the
syntactic material negated is not the entire clause (as it is in all cases of
standard negation) but only the object. Here, then, the syntactic field of
negation is not clausal but limited. Negation of a sentence function, however,
frequently leads to global negation, i.e. although the syntactic field of
negation is limited, the speaker presents and the hearer understands such a
sentence as negative as a whole.
As we saw in section 7.3, sentences describe situations with one or more
participants. In an S P O O sentence like They gave us the tickets the
participants involved are AGENT (the subject), BENEFICIARY (the indirect

object) and AFFECTED (the direct object). Now if one of these sentence
functions is negated, as it is in None of them gave us the tickets, They gave
none of us the tickets and They gave us no tickets, a participant is eliminated.
The situation described by the sentence as a whole therefore cannot
materialize, i.e. no situation arises where somebody hands over something to
somebody else. For that reason the hearer understands the sentence as a
whole to be negative. That there is global negation here appears from
application of the criteria, e.g. addition of a positive tag question:
(1) None of them gave us the tickets, did they?
(2) They gave none of us the tickets, did they?
(3) They gave us no tickets, did they?
Limited syntactic field of negation involves not only subjects and objects but
also complements and adverbials, e.g. She is no one of importance (is she?)
and At no point in the meeting did they tell us the truth (did they?).
Negation of a complement or of an adverbial, it should be added, is not
always accompanied by global negation. This can be illustrated by an S P C
sentence like Simpson pleads not guilty and an S P O A sentence like He'll
fix your car in no time. That the domain of negation is here local appears
from application of the criteria. For example, both sentences select a negative
tag question: doesn't he? /won't he?.
What we have shown in this section is that the syntactic field of negation is
either clausal (as it always is in the case of standard negation) or limited
(which it is if S, O, C or A is negated). In the former case the domain of
negation is always global. In the latter case it is very frequently global too.
7.5.5. Negative sentences with global domain and limited field

Global domain combined with limited syntactic field is found in sentences
with negative or negated quantifiers and pronouns, i.e. in sentences with no,
nothing, nobody, no one, none, neither or words like one, many, much, all
and every preceded by NOT:
(1) A predawn earthquake in eastern Java caused no casualties or serious
damage.
(2) My mother noticed nothing of this.
(3) Nobody's perfect.
(4) Neither of us said anything.
(5) Not one / Not many of the students will pass the exam.
(6) She is no one of importance.
Polarity 225
The syntactic field of negation is here limited to subjects, objects or com-

plements, but the sentences become globally negative all the same.
Occasionally global negation with negative quantifiers/pronouns is the
only alternative. Thus, for example, Nobody cares is the negative counterpart
of Somebody cares when somebody has non-specific reference (Somebody
doesn't care being possible only if Somebody has specific rather than non-
specific reference). In a number of idiomatic cases, furthermore, it is
practically the only alternative, and in existential sentences and sentences
with HAVE in the sense of 'possess' it tends to be preferred to standard
negation (see Johansson and Lysväg 1986: 241):
(7) It's no wonder.
(8) There's no claret in the decanter.
(9) We have no cash.
Otherwise, the choice between negation with a negative quantifier/pronoun
and standard negation is largely a matter of style. For example, the next two
examples can both be used to express the same kind of negative statement. It
should be added, though, that (11) differs from (10) in that it can also be used
as a denial, a use which is not likely in the case of (10):
(10) The earthquake caused no serious damage.
(11) The earthquake did not cause any serious damage.
Global quantifier negation may cooccur with (global) standard negation:
(12) Nobody didn't know the answer.
(13) Not many of the students didn't pass the exam.
Here both the negative elements are operative, so although (12) and (13) as a
whole are negative sentences - as application of the three criteria will show -
their meanings are roughly similar to those of Everybody knew the answer
and Most of the students passed the exam. This type of genuine 'double
negation' is also found in an example like They don't owe me nothing (=
They owe me something') and should be distinguished from the one found
outside Standard English, where there is negative concord (as in e.g.
Spanish) and where the extra negative adds nothing to the meaning of the
sentence apart from emphasis (They don't owe me nothing = They don't owe
me anything). As illustrated by / didn't say nothing to nobody, this type of
semantically redundant negation may even be 'multiple'.
We also find global negation in sentences with negative/negated adverbs:
(14) Alan Clark has never concealed his philandering ways.
(15) That approach will get you nowhere.
(16) Not everywhere are the effects of these policies noticeable.
In each example the syntactic field of negation is limited to the adverbial

function. But the sentence as a whole becomes negative.
Among the negative adverbs we should perhaps include BARELY, HARD-
LY, RARELY, SCARCELY and SELDOM. Formally, these words are not
marked as negative by an initial /n/, but semantically they can be paraphrased
as 'almost not1 (BARELY, HARDLY, SCARCELY) or 'not often' (RARELY,
SELDOM). The following examples illustrate a kind of 'global negation'
expressed in this way:
(17) I barely know her.
(18) Bill Clinton's reflexive faith in their efficacy is hardly surprising.
(19) She scarcely speaks a word of Danish.
In such examples, application of the tag question test and the NOT EVEN test
does not yield a completely clear result. In all three examples we can add a
positive tag (e.g. do I? in (17)), which indicates that they are negative
sentences. At the same time, however, they seem to prefer an EVEN clause to
a NOT EVEN clause (e.g. even to her husband in (19)), especially if there is
little or no pause between the two clauses. If constituting a separate
information unit, the NOT EVEN clause improves somewhat. Note finally
that all three examples select a coordinated clause with NEITHER or EITHER
instead of SO or TOO, so in this respect they are globally negative.
Before proceeding to non-global negation we must consider examples with
LITTLE or FEW like the following:
(20) The earthquake caused few casualties.
(21) The earthquake caused little damage.
In some descriptions of English grammar such sentences are considered
negative (unlike sentences with A FEW or A LITTLE). According to the tests
used here, however, they appear to be positive. Sentences like (20) and (21)
can be expanded with and so did the ensuing landslide, with and the ensuing
landslide caused few casualties / little damage too or with didn 't it?
Furthermore, they cannot be followed by NOT EVEN constructions or clauses
with NEITHER or EITHER, and the tag question will normally be didn't it?,
indicating that the matrix clause is positive.
Semantically, LITTLE and FEW resemble BARELY, HARDLY, SCARCELY
('almost not') and RARELY, SELDOM ('not often') in having negative
colouring ('not much', 'not many'). They also behave like these words with
respect to inversion. As mentioned in section 5.3.4, there is inversion in
sentences beginning with a negative or restrictive constituent other than the
subject (if negation/restriction applies to the whole sentence). There is thus
inversion in sentences beginning with explicitly negative words or with
Polarity 227
implicitly negative words like BARELY, HARDLY, etc. (as illustrated by

Hardly a word did he utter). However, there is also inversion in sentences
like the following beginning with LITTLE or FEW:
(22) Little did John expect this move.
(23) On few matters is our conception of the future this clear.
With respect to inversion, then, LITTLE and FEW behave like NEVER,
NOWHERE, etc. and BARELY, RARELY, etc. Furthermore, it is sometimes
difficult to apply the criteria to sentences beginning with these words. For
example, (22) cannot readily be followed by either and neither did his wife
or and so did his wife. It thus appears that sentences with LITTLE or FEW are
not entirely determinate with respect to polarity, i.e. that there is a grey zone
between clearly positive and clearly negative sentences. In this grey zone -
though closer to the negative end of the scale - we should probably also
place sentences with BARELY, RARELY, etc.
7.5.6. Local negation

The most obvious cases of local negation are those where a word begins with
a negative prefix and where negation operates within the limits of the word.
In a sentence with a negative prefix like the following there is thus no global
negation:
(1) These distinctions were virtually nonexistent before the mid-17th century.
As illustrated by the next example, however, a word beginning with a neg-
ative prefix may naturally also occur in a sentence with global negation:
(2) This situation is not atypical.
As already shown, local negation outside the boundaries of the word may be
expressed by NOT. This is common before adjectives or adverbs, particularly
if these begin with a negative prefix:
(3) Our new therapy is a not insignificant step in the right direction.
(4) Not infrequently there is torrential rain on this part of the coast.
(5) Simpson pleads not guilty.
(6) Not long ago Marion turned eighty.
The domain of negation is also local in sentences like / hope not and I'm
afraid not in which NOT realizes an object and the complementation of an
adjective respectively and is used as an anaphor with clausal antecedent, for
example in reply to a question like Are Norway winning?
Local negation with negative elements other than NOT or prefixes can be
illustrated by examples like these:
(7) We'll have your roof fixed in no time.

(8) A hundred pounds for that room is nothing short of robbery.
(9) She had a barely noticeable smile on her face.
The different types of negation examined so far are summarized and exem-
plified in table 1:
DOMAIN FIELD
clausal γοιι can>t come tomorrow
(standard)
global
limited:
a) S No man is an island
b) Od My mother noticed nothing
Negation
c) Oi They gave none of us the tickets
d) C She's no one of importance
e) A At no point did they surrender
local: a) negative prefixes This clause is nonrestrictive

Not long ago Marion turned eighty
c) other negative elements We'll have your roof fixed in no time
Table 1: Types of negation
7.5.7. Nonassertive forms

Some English words are typically restricted to occurring in negative
sentences, interrogative sentences and conditional clauses, i.e. in con-
structions which do not express an assertion the way positive declarative
sentences usually do. One such nonassertive form is EVER:
(1) She hasn't ever made me happy.
(2) Has she ever made you happy?
(3) If I have ever made you happy I will not have lived in vain.
(4) *She has ever made me happy.
EVER also occurs in comparative clauses, which resemble conditional
clauses in not expressing an assertion in themselves:
(5) She is making me happier than I have ever been before.
Forms typically limited to nonassertive contexts in English are any, complex
pronouns and adverbs beginning with any (anything, anybody, anyone,
Polarity 229
anywhere), either, ever, yet and at all. The corresponding assertive forms are
some (something, somebody, someone, somewhere), too, sometimes and
already (at all having no obvious counterpart). For example, the assertive
version of (1) is She has sometimes made me happy. In negative sentences,
the use of nonassertive forms can be illustrated by the following examples:
(6) My father had given no hint to her of any malaise.
(7) To my surprise, I didn't see anybody/anyone.
(8) My father has never liked bridge either.
(9) The Prime Minister hasn't resigned yet.
(10) They've done nothing at all about it.
As can be seen, nonassertive forms occur both in sentences with standard
negation and sentences with other types of global negation. Sometimes we
also find them in positive declarative sentences in which there are words
with negative prefixes (local negation) or negative meaning:
(11) We disapprove of any intervention in that area.
(12) We oppose any intervention in that area.
The use of nonassertive forms in implicitly negative sentences with the
adverbials BARELY, HARDLY, RARELY, SCARCELY, SELDOM is illustrated
by the next examples:
(13) They have barely settled down yet.
(14) Bill Clinton's reflexive faith in then- efficacy is hardly surprising either.
(15) They rarely go anywhere outside Britain.
In all the examples given above, the nonassertive form follows the negative
element. This order cannot be reversed. While we can say She hasn 't ever
kissed anyone we cannot say *She has ever kissed no one.
With respect to assertive vs. nonassertive forms, LITTLE and FEW behave
like explicitly and implicitly negative elements:
(16) The earthquake caused little damage anywhere.
(17) Few people have ever made me that happy.
ONLY, which resembles LITTLE and FEW as well as BARELY, HARDLY,
etc. in having a negative hue ('with no others in the same class'), also
combines with nonassertive forms:
(18) Only my husband has ever made me that happy.
For further discussion of nonassertive forms and examples like Anybody can
do this and Did you see someone in the garden?, the reader is referred to
section 11.3.4 [A.fj.
7.5.8. Semantic scope of negation

We have so far discussed the overall polarity of clauses in connection with
negation (domain of negation) and the syntactic material actually negated
(syntactic field of negation). In this section we take a closer look at what
parts of a clause are affected semantically by negation (semantic scope of
negation). Consider first examples like the following:
(1) Fortunately the earthquake didn't cause casualties.
(2) Fortunately the earthquake caused no casualties.
In both of these the domain of negation is global. On the other hand the two
examples differ in that the syntactic field of negation is clausal in (1) and
limited (to the object) in (2).
In order to give a full account of negation in these examples, it is not
enough to account for domain and field: we must also specify that negation -
though global - has no semantic influence over the part of the meaning
referred to by fortunately. To cope with examples of this type we need the
concept of semantic scope. The semantic scope of negation may be either
complete or incomplete. While it is incomplete in examples like (1) and (2)
in that the meaning described by fortunately is unaffected by negation, it is
complete in examples like the next ones:
(3) At first the earthquake didn't cause casualties.
(4) At first the earthquake caused no casualties.
Here, then, the part of the meaning described by at first is inside the semantic
scope of negation.
In an example like (2) the domain, syntactic field and semantic scope of
negation are all different: the domain is the entire clause, the syntactic field is
the object and the semantic scope is the part of the meaning described by
everything except fortunately. While field is exclusively syntactic (involving
the negation of syntactic entities like S, O and clause) and scope is ex-
clusively semantic (involving the effect of negation on meaning), domain is
both syntactic and semantic: it is established on syntactic criteria (cf. the tests
mentioned in section 7.5.3) and relates to the clause; at the same time,
however, it concerns the language user's conception of a sentence as negative
or positive on the most general level.
In examples where the domain of negation is local, domain, syntactic field
and semantic scope all go together. This can be illustrated by an example like
'Simpson pleads not guilty', which has local domain, limited field and
incomplete scope.
In the case of global negation, the negative may have semantic scope over
the entire simple sentence. This is true of (3) and (4) above and for an
Polarity 231
example like Sanctions don't challenge vital interests, which can be analysed
semantically as [Not [sanctions challenge vital interests]]. That it is not just
the subpart of meaning following n't that lies within the semantic scope of
negation, but also the subject noun, is evident from the fact that the sentence
may be produced with contrastive stress on sanctions, i.e. the focus of
negation may be on this word (see section 7.5.1). Not infrequently, however,
some of the semantic material encoded in a sentence with global negation
falls outside the semantic scope of negation.
This is often the case with adverbials (as illustrated by examples (1) and
(2)). In the first place, disjuncts and conjuncts always behave in this way:
(5) Sanctions don't challenge vital interests, unfortunately.
(6) Sanctions don't challenge vital interests, however.
These examples can be paraphrased 'Unfortunately/However it is not the case
that sanctions challenge vital interests', i.e. they must be analysed
semantically as [Unfortunately/However [not [sanctions challenge vital
interests]]]. Adjuncts differ from disjuncts and conjuncts in that, more often
than not, they fall within the semantic scope of negation. This is always the
case with obligatory adjuncts:
(7) Software is not like other intellectual property.
(8) He didn't place the figures in the right order.
Optional adjuncts falling inside the scope of negation can be illustrated by:
(9) Without autonomy, we cannot have a truly free and democratic society.
(10) This bottle isn't labelled very clearly.
(11) Nothing much had changed in 1994.
In some cases, however, optional adjuncts fall outside the semantic scope of
negation. This is commonly the case with adverbials of reason or purpose:
(12) For those reasons, I didn't write the book.
(13) With this objective in mind, we have neglected none of the suggestions from
our customers.
If such an adverbial is placed finally, it may also stand outside the semantic
scope of negation. In an example like / didn't -write the book for those
reasons, the scope is the same as in (12) if the adjunct is pronounced without
prosodic prominence:
(14) I didn't write the 'book for those reasons.
If the adjunct is pronounced with prosodic prominence, however, it is
brought within the semantic scope of negation. Nuclear stress may be placed
either on reasons or - contrastively - on those, and the tonal contour is

typically falling-rising:
(15) I didn't write the book for those vreasons.
(16) I didn't write the book for vthose reasons.
Whether an optional adjunct is inside or outside the semantic scope of
negation may be signalled not only prosodically but also by constituent
order. In an example like / didn't kill him deliberately where it follows the
negative elements, the adjunct is inside the scope of negation. But in
Deliberately, I didn't kill him and / deliberately didn't kill him, where the
adjunct precedes the negative, the negative has no semantic influence over it.
If an adjunct is outside the semantic scope of negation, the sentence which
remains if this adjunct is removed is entailed by the sentence in which it is
retained (see Huddleston 1984: 429). What this means becomes clear when
we look at an example like I didn't kill him which is entailed by Deliberately,
I didn't kill him and I deliberately didn't kill him. On the other hand, it is not
entailed by 7 didn 't kill him deliberately, where the adjunct is inside the
semantic scope of negation. What this sentence implies is that the person
referred to by him was in fact killed. If in doubt about the role played by an
optional adverbial with respect to scope of negation, we can thus use the test
of entailment. This test will show the adjunct to be outside the scope of
negation in examples like (12) and (13) (entailment) but inside the scope of
negation in examples like (9) through (11) (no entailment).
In addition to adverbiale, some modal auxiliaries may stand outside the
semantic scope of negation. As illustrated by the following example, this
goes for MAY used in the sense of 'be perhaps likely to':
(17) She may not understand your decision.
This sentence can be paraphrased as 'It is possible that she won't understand
your decision' and analysed as [Possible [not [she understand your
decision]]]. If MAY is used in the sense of'be allowed to', it normally stands
inside the scope of negation. For example, a sentence like You may not
borrow my car can be paraphrased as 'You are not permitted to borrow my
car' and analysed semantically as [Not [permitted [you borrow my car]]]. But
as illustrated by This once you may not do the dishes, used in the sense of Ί
permit you to refrain from washing up', it may also stand outside the scope of
negation: [Permitted [not [you wash up]]]. For further discussion of negation
in connection with modal auxiliaries, see section 9.9.
Note finally examples with indefinite quantified subjects like the
following:
(18) Many arrows didn't hit the target.
Polarity 233
(19) Two goals weren't scored by Michael Laudrup.

There is a sense in which the subject is here excluded from the semantic
scope of negation. Thus the fact that many arrows didn't hit the target does
not exclude the possibility that many (other) arrows did hit the target.
Similarly, the fact that two goals were not scored by Michael Laudrup does
not exclude the possibility that Michael Laudrup did score two (other) goals.
Negation is without a doubt a highly complicated field of grammar. In
tackling the analysis of negation in specific sentences, the following three-
stage approach is recommended:
(i) Find out whether the domain of negation is global or local by using the
criteria mentioned in section 7.5.3.
(ii) Then find out whether the syntactic field of negation is limited to a
specific sentence function or is clausal.
(iii) Finally find out whether the semantic scope of negation is complete or
incomplete by examining whether all the semantic material encoded in the
sentence is influenced by negation or not.
When applied to a sentence like Fortunately the earthquake caused no
casualties, this approach will demonstrate that the domain of negation is
global, that the syntactic field of negation is limited to the object and that the
semantic scope of negation is incomplete in that the meaning expressed by
the first word is unaffected by negation.
Here is an overview of the possible combinations of domain, field and
scope:
global unlimited complete

domain syntactic semantic
field scope examples
+ + + The earthquake didn't cause casualties
: + - Fortunately the earthquake didn't cause casualties
+ - + The earthquake caused no casualties
+ - - Fortunately the earthquake caused no casualties
- - - Simpson pleads not guilty

Graphic representation:
The earthquake didn't cause casualties

scope l 1
field l l
domain ' '
Fortunately the earthquake didn't cause casualties

scope l-1
field I_
domain I_ '
The earthquake caused no casualties

scope l-1
field l -1
domain '-'
Fortunately the earthquake caused no casualties

scope I--1
field I_ I
domain I_ I
Simpson pleads not guilty

scope l-I
field I-1
domain '-'
7.5.9. Non-declarative sentences

In yes-no interrogative sentences there is negation if the speaker's expecta-
tion with respect to positive-negative is not neutral but positive:
(1) Isn't your attitude changing?
(2) Didn't they prove uncooperative?
Here the speaker seeks information but at the same time expresses that he
expects the state of affairs referred to to hold (Tell me if I'm right in
assuming that your attitude is changing/that they proved uncooperative1). In
WA-interrogative sentences the speaker's assumption is always positive. A
positive sentence like Who persuaded you? requests information about the
identity of the person who performed an action the speaker assumes to have
taken place ('Somebody persuaded you. Who was it?'). Here the negative
Polarity 235
counterpart Who didn't persuade you? is not very natural, but if it is made
possible by contextualization it clearly does not refer to the same situation as
the positive sentence. In some cases a wA-interrogative sentence has no
negative counterpart, e.g. Whatever put that idea into your head? On the
other hand, negative sentences with why are very common:
(3) Why don't you live in London?
(4) Why didn't it work?
Semantically, these sentences are obviously quite different from the
corresponding positive sentences. While (3) can be paraphrased as 'Tell me
the reason you don't live in London', a positive sentence like Why do you live
in London? can be paraphrased as Tell me the reason you live in London'.
Negative imperative sentences are formed by placing don't before the main
verb, also when this is BE:
(5) Don't knock him unconscious.
(6) Don't be afraid.
Note in passing that BE only accepts DO-support in negative imperatives like
(6) and in emphatic imperatives like the following:
(7) Do be nice to him.
A >>ott-subject may be used in imperative sentences to add an element of e.g.
displeasure. This is possible in negative imperative sentences as well:
(8) Don't you knock him unconscious.
Here the presence of you makes the message come across as a threat.
Exclamatory sentences differ from other sentence types in not normally
being capable of negation. For example *Ho\v beautiful she doesn't look! is
not grammatical in any context. On the other hand a negative exclamatory
sentence like How cleverly she doesn 't stop speaking! cannot be ruled out in
a context where the person talked about tries to retain the upper hand in a
debate. The illocutionary value 'exclamation' is often expressed by negative
sentences, but these nearly always belong to other sentence types,
particularly the interrogative one. This can be illustrated by examples like
Wasn't she lovely! and Isn't he English!, the communicative function of
which, despite the interrogative form, is not to seek information but to
indicate an emotional reaction.
7.5.10. Emphasis and focus

The dummy operator DO is used not only in connection with negation and
inversion but also to create emphasis:
(1) Sanctions do challenge vital interests.

In this example the operator DO - pronounced with nuclear stress - is used to
emphasize the truth of the sentence. What is emphasized in this way is the
entire statement. If a speaker wishes to underline only individual parts of a
sentence, he will do so by means of strong nuclear stress exclusively (see
section 7.5.1 on focus of negation). For example, a sentence like Jane speaks
English correctly may be emphasized not only as a whole by means of DO
but also in part by focussing on a constituent prosodically, i.e. by
pronouncing its realization with strong nuclear stress:
(2) 'Jane speaks English correctly, (but not Tarzan)
(3) Jane * speaks English correctly, (but spells it incorrectly)
(4) Jane speaks 'English correctly, (but not French)
(5) Jane speaks English cor'rectly. (but not fluently)
In these examples the special attention given to an individual constituent by
means of prosody is contrastive (so we speak of 'contrastive focus' here).
When DO is used in positive declarative sentences, it is often used not only
to emphasize the entire statement but simultaneously to express a contrast to
the corresponding negative statement, i.e. a kind of'denial in reverse':
(6) A: Why didn't you own up?
B: I "did own up.
In using Jane does speak English correctly the speaker may express a
contrast to something mentioned before too, but he can also use it for
emphasis exclusively, for example to signal that Jane's command of English
strikes him as remarkably impressive. The following examples with DO
illustrate emphasis which is not necessarily contrastive:
(7) Lucky girl! But I am relieved. I do worry, you know.
(8) I won't be alone. But I do worry about you here in the flat.
Expansion by means of DO for the expression of emphasis is found not only
in positive declarative sentences but also in positive imperative sentences:
(9) Jane, do ask Harry to join us. (complex sentence)
When a speaker wishes to emphasize the entire statement expressed by a
sentence in which the predicator is realized by a group with one or more
auxiliaries, she will do so by pronouncing the operator with nuclear stress:
(10) Accidents 'will happen.
(11) A: Why haven't you owned up?
B: I 'have owned up.
Polarity 237
In emphatic sentences with or without DO the part following the operator is

often contextually redundant and may therefore be omitted (see section 4.2.3
on ellipsis and pro-forms):
(12) A: Why didn't you own up?
B: I did.
(13) A: You weren't expecting to do any work, were you?
B: I was, actually.
As the auxiliary HAVE and the modal auxiliaries are used for the expression
of time and modality respectively, sentences with these verbs are frequently
used to draw attention to temporal contrasts or to contrasts between real and
possible or desirable states of affairs (see Quirk et al. 1985: 1371f):
(14) A: They are a handsome couple, aren't they?
B: They have been, no doubt. A little past their best now, perhaps.
(15) A: I'd like to treat us both.
B: You could, of course.
(16) A: Well, I think they enjoyed it.
B: They should have done. It was not exactly an inexpensive evening.
In negative sentences, denial of the entire statement expressed by the
corresponding positive sentences is emphasized by assigning nuclear stress
either to NOT or - in the case of NOT-contraction - to the operator (see
sections 7.5.1-2):
(17) Jane does 'not speak English correctly.
(18) Jane * doesn't speak English correctly.
(19) Jane can ' not speak English correctly.
(20) Jane 'can't (or 'cannot) speak English correctly.
Under neutral conditions nuclear stress falls on the last content word (as
mentioned in section 7.5.1). In an example like (5) the prosodic prominence
given to the word correctly therefore need not signal contrastive focus but
may signal ordinary end-focus instead. In closing this section we should add
that nuclear stress may also signal corrective focus. In an example like She
wound a bandage round my foot pronounced with nuclear stress on the verb
the speaker's purpose in selecting this prosodic pattern may be to correct a
misunderstanding on the part of her interlocutor, for example to make it clear
that the verb form she had used was not bound.
7.6. Subject-predicator concord

7.6.1. Introduction
By concord is understood agreement in form between different linguistic
units. In noun groups like this girl and these girls, for example, the pronoun
and the noun agree with respect to number: in the former both forms are
singular and in the latter they are plural. Concord involves not only number
but also other grammatical categories. In a sentence like I'm sorry there is
agreement between the pronoun realizing the subject and the verb realizing
the predicator not only with respect to number (singular) but also with
respect to person (first). In a language like French there is concord of gender
in noun groups like un petit signe 'a small sign' and une petite clef a small
key'. Here the forms of the indefinite article and the adjective agree in being
masculine in the first group and feminine in the second, due to the fact that
the gender of signe is masculine and that of clef feminine. Though gender is
not an inherent feature of nouns in English, a related kind of agreement is
found in examples like This is my son, who lives in London and This car,
which is brand-new, belongs to Ann where the choice of relative pronoun is
determined by the personal vs. nonpersonal meaning of the preceding noun.
Concord may be internal in the sense that the co-occurrence restrictions it
imposes involve the realization of group constituents (as in this girl/these
girls) or external in that it involves the realization of sentence functions (e.g.
of subject and predicator in I'm sorry).
7.6.2. The basic rule

In English there is agreement between subject and predicator when the latter
contains a present form of a full verb or of one of the primary verbs BE,
HAVE and DO. The -s form of the finite verb is used if the subject is realized
by a singular nominal or third person singular pronominal:
(1) France has always relished its "special role" in Africa.
(2) She doesn't know much about Rwanda.
(3) This consideration has led France to play gendarme in Africa.
(4) The stated objective seems laudable.
Otherwise the -0 form of the verb is used:
(5) France and England have always disagreed on this point.
(6) They don't know much about Rwanda.
(7) These considerations have led France to play gendarme in Africa.
(8) The stated objectives seem laudable.
Subject-predicator concord 239
This basic rule does not apply to sentences with modal auxiliaries (e.g. The
stated objective(s) may seem laudable) or to sentences or clauses in the
subjunctive mood (e.g. Long live Prince Charles / I suggest he see a
psychiatrist). In those cases where the predicator is realized by a form of BE,
on the other hand, concord involves not only the present form but also the
past form: if the subject is realized by a singular nominal, the first person
singular pronoun / or a third person singular pronominal, the verb form
selected is was (e.g. Escape was easier than before / My mother was
concerned about her brother's health /1 was clearly an embarrassment on
that visit / Nothing of this was said to me). Otherwise the form selected is
were. In the present form of BE, furthermore, there are not two but three
distinctions: am is selected if the subject realization is /, is if it is in the third
person singular and are otherwise. In the following account of concord, are
and were are to be regarded as the present and past -0 form of BE.
Subject-predicator concord is largely a matter of inflectional co-variation
(see Huddleston 1984: 241). If the nominal realizing the subject is in the
singular, the verb adds the inflection -s (as illustrated by My sister lives in
Rome). If the subject nominal is in the plural, however, and thus in regular
cases marked by the inflection -s on the noun, no suffix is added to the verb
(e.g. My sisters live in Rome). As the relation between noun form and verb
form is thus of an automatic nature, it is not surprising that the meaning of
the verb in sentences in the present form is highly redundant with respect to
number. In an example like (4), for instance, the suffix -s only reflects the
singular meaning of objective and thus adds very little to the meaning of the
sentence (see Juul and S0rensen 1978: 13f). In (2), similarly, the suffix -s in
doesn't know only reduplicates the meaning of she with respect to person and
number. Verb inflection (-s vs. -0) is not invariably bleached, though. In
sentences like Counsel representing the payees expects a settlement and
Counsel representing the payees expect a settlement, it obviously makes a
great difference that the suffix -5 is present and absent respectively, for here
the head of the subject group is realized by a noun which shows no inflec-
tional difference between singular and plural.
Number is essentially a category of nouns and of pronouns (cf. distinctions
like book : books and this : these), and person is essentially a category of
pronouns (e.g. /: you : it). The reason we say that a verb is, for example,
singular and third person is that the -s which accompanies it (e.g. becomes /
has become / is becoming) is selected if the noun realizing the (head of the)
subject (group) is in the singular (e.g. tobacco I my sister) or the pronoun
realizing the (head of the) subject (group) is in the third person singular (e.g.
she /neither of us).
Though the basic rule of concord is quite simple, concord problems arise for
three main reasons (see Johansson and Lysvag 1987: 346ff): 1) in some cases
it is not obvious whether the form realizing the subject or head of the subject
group is singular or plural; 2) syntactic concord may be overruled by so-
called notional concord; 3) the rule of syntactic concord may be broken,
particularly in long sentences, because there are nouns between the subject
head noun and the predicator whose number differs from that of the subject
realization (a side-tracking factor referred to as 'attraction').
7.6.3. Singular or plural subject realization?

Normally, the noun realizing the (head of the) subject (group) is marked
inflectionally as singular or plural by means of -s and -0 respectively. In
some nouns, however, there is no such inflectional distinction. In order to
determine whether the subject realization is in such cases singular or plural,
it is helpful to look at determiners, certain pronouns and the noun realizing a
subject complement or the head of a subject complement noun group. As
shown in section 7.6.2, some nouns do not inflect for number. However, their
number can often be determined as singular or plural in the linguistic
context:
(1) The sheep ignore/ignores the dogs.
(2) This/These sheep will die soon.
(3) The sheep will die unless you give it/them an injection.
(4) Jane's sheep had been an ideal pet.
(5) In the mist the sheep looked like mole-hills.
Under those conditions where the rule of concord does not apply, and where
there are no number-revealing determiners, anaphoric pronouns, subject
complement realizations, etc., the number of the subject noun can only be
determined by looking at the larger linguistic context or the situational
context (cf. Juul and S0rensen 1978: 55ff). An example illustrating this
situation is The sheep may die before the vet gets here, which is ambiguous
as it stands. Other nouns behaving like sheep are counsel, craft, deer, cod,
means, (gas)works, headquarters, series, etc., see section 10.4.2 below.
Let us now investigate subject-predicator concord in a number of other
cases where the subject realization is not clearly marked as either singular or
plural:
A) Subjects realized by number-invariable nouns. In English some nouns
are restricted with respect to singular/plural contrast. Such nouns, which are
often referred to as number-invariable nouns because they are invariably
singular or plural, are relevant in any discussion of subject-predicator
concord. We distinguish the following general types (for more details, see
section 10.4.3 on the number category):
(i) Nouns with 'plural form' used as singular nouns only, such as names of
sciences and subjects like mathematics ana phonetics, diseases like measles
and mumps, of games like billiards and darts, and the individual mass noun
news. These select the -s form of the verb.
(ii) Nouns with 'singular form' used as plural nouns only, such as nouns
referring to a collection of entities like cattle, police, clergy, livestock,
vermin, etc. These select the -0 form of the verb.
(iii) Nouns with plural form only used as plural nouns, such as names of
tools and articles of dress consisting of two parts like scissors, shears,
binoculars ana jeans, tights, pyjamas and some geographical proper nouns,
for example, the Andes, the Alps, the Hebrides. These select the -0 form of
the verb. Note also nouns like customs, contents, colours, pains, etc., which
have semantically distinct singular forms (custom, content, colour, pain).
(iv) Nouns with singular form only used as singular nouns, such as mass
nouns like anger, furniture, music, peace, poverty, resentment, etc. and
proper nouns (John, Rome, Spain, etc.). These select the -s form of the verb.
B) Subjects realized by compound units. If the conjoints are plural, the -0
form is selected: The boys and girls are now with their parents / The boys or
the girls are on duty now. If the conjunction is and and the conjoints are
realized by singular nominals, the compound unit selects the -0 form of the
verb, i.e. counts as plural: Bradbury and his partner bear a heavy
responsibility. If the conjunction is or, on the other hand, concord is usually
determined by the last conjoint: Bradbury or his partner bears a heavy
responsibility / Bradbury or his partners are likely to want a settlement / His
sisters or his father is to blame. One exception to this rule is the expression
one or more followed by an o/-construction containing a plural noun, cf.:
One or more of these conditions has not been observed vs. One or more
politicians have leaked information to the company. Constructions with the
correlative conjunctions both ... and and either ...or behave like construc-
tions with and and or respectively, i.e. insertion of both in the first of the
examples just given and of either in the second would not affect the choice
of verb form. In those cases where one or both of the conjoints of a
compound unit are realized by a singular pronoun, concord operates in the
manner just outlined: there is plural concord if the conjunction is and (She
and Bradbury bear a heavy responsibility) and singular concord if it is or
(She or Bradbury bears a heavy responsibility).
The correlative conjunction neither... nor is the negative counterpart of both

... and and has inclusive meaning. Nevertheless, a compound unit in which
the conjoints are realized by singular words or groups has singular concord
when coordinated by neither... nor: Neither Bradbury nor his partner bears
a heavy responsibility. In informal language plural concord cannot be ruled
out here.
Special problems arise where the conjoints of a construction with the
coordinators or and either ...or are realized by nominals or pronominals
which differ with respect to number or person. In such cases there is some
vacillation, but most commonly it is the -0 form of the verb which is
selected (see also section 7.6.5):
(6a) (Either) the cabinet ministers or Major himself refuse to face reality.
(7a) (Either) you or your partner are crazy.
However, there is sometimes 'attraction' to the last conjoint:
(6b) (Either) the cabinet ministers or Major himself refuses to face reality.
(7b) (Either) you or your partner is cra2y.
Subject compound units with pronominal conjoints in different persons are
often felt to be awkward. When they do occur, the -0 form is normally used,
though attraction to a third-person pronoun cannot be entirely ruled out:
(8a) (Either) you or he are crazy.
(8b) (Either) you or he is crazy.
(9a) (Either) they or I are crazy.
(10) (Either) she or I are crazy.
There is no attraction to a first-person pronoun:
(9b) *(Either) they or I am crazy.
If the subject is realized by a group in which the noun is premodified by a
coordinated adjective construction with and, the -0 form of the verb is
selected if the speaker has a plurality in mind:
(11) Primary and secondary education require more skilled teachers.
If the speaker does not regard the situation in this way, however, she will
select the -s form of the verb:
(12) Secondary and tertiary education needs support from the government.
C) Subjects realized by clauses. In complex sentences the subject is
sometimes realized by a finite or nonfmite clause. As already indicated, in
such cases it is the -s form of the verb which is selected, i.e. clausal subjects
count as singular:
(13) That I have done a thing like that bothers me night and day.
(14) To have done a thing like that bothers me night and day.
The singular nature of such clausal subjects appears from the fact that they
can be replaced by the singular pronoun it (It bothers me night and day). An
exception to this rule is constituted by independent relative clauses, i.e.
clauses beginning with a wA-word which functions both as relative and
antecedent (see sections 6.4.2 and 11.3.3), but then, their status as clauses
rather than groups is not unproblematic. If what or whatever determines a
plural noun in such a subject clause, the predicator of the matrix clause is
always realized by the -0 form of the verb:
(15) What friends he has live abroad.
(16) Whatever guests you invite are welcome.
In S P C sentences with a wAaf-clause as subject and in which the
complement is realized by a plural noun or noun group, there tends to be
plural rather than singular subject-predicator concord:
(17) What is required now are (is) food, drink and good company.
More generally, independent relative clauses resemble noun groups or
pronoun groups. For example, What friends he has ... can be paraphrased as
The friends he has ... or Such friends as he has ... and What is required now
... as That which is required now ... . It is therefore not surprising that such
clauses behave differently from others with respect to concord.
Another exception concerns subjects realized by discontinuous 'nominative
with infinitive* clauses, i.e. constructions with raised subjects (see section
5.6.1). In sentences like They are believed to be guilty and The children
appear to be hungry where the subject of the /o-infmitive is realized by a
plural form, it is the -0 form of the finite verb which is selected in spite of
the fact that its subject is realized clausally (They ... to be guilty / The
children ... to be hungry).
D) Subjects realized by pronouns. If a subject is realized by an indefinite
pronoun, there is in most cases singular concord. For example, somebody,
anybody, everybody, nobody and the corresponding pronouns ending in -one
select the -s form of the verb, as illustrated by No one knows and Everybody
loves somebody sometime. Only in those cases where a pronoun from this
group is postmodified by a preposition group in which the dependent is
realized by a plural form do we occasionally find plural concord as a result of
attraction (cf. section 7.6.5): Nobody except members of the Conservative
Party believe(s) this. Pronouns like one, each, something, anything and
everything naturally enough take singular concord, too.
Some indefinite pronouns behave differently, however. Either and neither

prefer singular concord, for example in (N)either seems qualified for the job,
but when postmodified by an o/-group in which the dependent is realized by
a plural construction, it is often the -0 form of the verb which is selected,
particularly in the case of neither (Neither of the applicants seem qualified).
With non-textual reference to persons, none prefers plural concord, as in
None are so deaf as those that will not hear. When postmodified by a
preposition group containing a plural (pro)noun - as it typically is - none
selects the -5 form of the verb in formal BrE and the -0 form in American
and informal BrE. This is the case both when the pronoun group refers to
persons (e.g. None of the applicants seem(s) qualified) and to a plurality of
nonpersonal entities (e.g. None of your suggestions seem(s) useful). Under
these conditions some and all always have plural concord (Some/All of the
applicants seem qualified; Some/All of your suggestions seem useful), and
when these pronouns realize a subject on their own, plural concord is
obligatory in the case of personal reference too (Some like it hot /All accept
your proposal). With reference to something non-countable none, some and
all select the -s form, e.g. Some (of the cheese) has been left in the fridge.
When all means 'everything' it takes the -s form, even if a plurality is implied
(e.g. All is lost now /All we need now is doctors).
When realizing a subject, the interrogative pronouns what, which and who
take singular concord if the speaker refers to a singular entity and plural
concord if she refers to a plurality. This can be illustrated by examples like
What is your reason vs. What are your reasons? / Which (of the two books)
is the more interesting vs. Which (of these books) are in stock? I Who is that
girl over there? / Who are the girls over there? When independent relative
what is used as a subject in an independent relative clause (cf. section
11.3.3), it usually takes singular concord but may occasionally take plural
concord if the speaker has a clear plurality in mind:
(18) What is (are) required now are food, drink and good company.
A provisional subject realized by the personal pronoun it always selects the
-s form of the verb, as in It seems likely that they won't come and It seems
advisable to accept their offer. As the real subject in such cases is realized by
a finite or nonfmite clause, and as clauses count as singular with respect to
concord, this is hardly surprising. On the other hand, concord in sentences
with a provisional subject realized by there is determined by the number of
the form realizing the real subject. This can be illustrated by examples like
There's a fly in my soup and There are good reasons for this. In colloquial
language, however, there is singular concord also in those cases where the
real subject is realized by a plural form (e.g. There's many Danes who dislike
European integration). If the real subject contains the expression more than
one or the coordinator both ... and with a singular first conjoint, there is
singular concord (despite the plurality of the real subject as a whole) even in
formal language: There is/*are more than one reason for this / In this
construction there is/*are both an adverbial and an object.
7.6.4. Notional concord

Syntactic subject-predicator concord is sometimes overruled by notional
concord. This can be illustrated by the following examples:
(1) The department are now in full agreement.
(2) The first six months was spent in India.
In (1) the singular noun of the subject group combines with the plural form
of BE in order to describe the referent of this noun group as a plurality, i.e. to
focus on the separate members of a set rather than on the set as a unit. In (2),
conversely, the plural noun of the subject group combines with the singular
(past tense) form of BE in order to describe the referent of this group as a
unit, i.e. as a temporal whole rather than six separate sections of time. In both
examples there is thus syntactic discord but notional concord. What they
illustrate is that English has a certain degree of freedom in interpreting the
referent of a noun group as a unity or a plurality, irrespective of the
grammatical number of the noun in this group.
Let us now turn to the specific conditions under which notional concord
prevails and the rule of syntactic concord is breached.
A) Collective nouns. A noun like DEPARTMENT obviously inflects for
number (department: departments), but when its singular form realizes the
head of a subject group it does not invariably select the -s form of the verb.
As illustrated by example (1), it may for semantic reasons select the -0 form
instead. DEPARTMENT is a so-called collective noun, i.e. a number-
inflecting noun denoting a set of members and which in the singular is
capable of combining not only with the -s form but also with the -0 form.
Other nouns belonging to this class are AUDIENCE, BAND, CHORUS,
CLASS, COMMITTEE, COMPANY, CROWD, FAMILY, FIRM, GOVERN-
MENT, HERD, JURY, MAJORITY, OPPOSITION, etc.
The Janus-like nature of a collective noun in the singular form is reflected
not only in the form of the verb but also in the selection of coreferential
pronouns, whether personal, possessive, reflexive or relative (e.g. it vs. they,
its vs. their, itself vs. themselves and which/that vs. who). In the following
examples with syntactic discord it is the forms of such pronouns indicating
plural or - in the case of the relative - personal antecedent which have been
selected together with the plural verb forms:
(3) The orchestra are playing poorly tonight, for they haven't been rehearsing.
(4) The committee are clearly pleased with their progress.
(5) The jury seem to be disagreeing among themselves.
Sometimes the coreferential pronouns they and their are used instead of it
and its even in those cases where a collective noun selects the -s form:
(6) The company needs a new managing director for their Paris branch.
Conversely, the relative pronoun selected may be which or that rather than
who in those cases where the collective noun group selects the -0 form:
(7) The crowd which/that has gathered at the entrance keep shouting slogans and
throwing stones.
The 'good language' requires consistency across all the relevant elements in
the sentence, but inconsistency (as in (6) and (7)) is by no means uncommon.
In AmE, the -s form of the verb is practically always selected if the noun
group contains a collective noun in the singular (see Preisler 1992: 163). The
referent of such a noun group is thus not usually interpreted as a plurality in
this variety of English. However, there may be conflict of number between a
singular verb form and a following pronoun which has the same reference as
the noun group:
(8) The band is playing poorly tonight, aren't they?
In a weak sense, then, the Janus-like nature of collective nouns may show
itself syntactically even in AmE.
The collective singular vs. collective plural treatment of a collective noun
is affected by certain syntactic factors. If such a noun combines with singular
determiners like a, this, that, (n)either and each, it nearly always selects the
-s form of the verb:
(9) This commission recommends the abolition of entrance exams.
If a collective noun combines with number-neutral determiners like the, my,
your, Peter's, on the other hand, the determiner does not prevent a collective
plural interpretation of the referent. Secondly, the presence of a subject
complement realized by a plural noun, or by a noun group in which the head
is realized by a plural noun, rules out selection of the -5 form of the verb,
even in AmE:
(10) The maj ority are members.
(11) My family are early risers.
Proper nouns referring to companies, organizations, etc. like Ford, Leyland

and UNESCO are, strictly speaking, not collective nouns in the sense
described above as they do not inflect for number. However, in BrE, they
resemble such nouns in permitting selection of the -0 form of the verb for
the purpose of expressing collective plural meaning (e.g. Toyota have
decided to launch yet another advertising campaign). Singular names of
sports teams very occasionally select the -s form of the verb, but in an
overwhelming majority of cases they take plural concord, as in Arsenal are
playing Tottenham on Saturday (in AmE only: Arsenal is playing Tottenham
on Saturday). With respect to subject-predicator concord they therefore
basically behave like the unmarked plural nouns CATTLE, POLICE, etc.
discussed in section 7.6.3.
B) Plural expressions of quantity and measure. In example (2) above -
The first six months was spent in India - we saw that there may be syntactic
discord in order to obtain the opposite effect from the one obtained by
combining a collective noun in the singular with the -0 form of the verb.
Here the rule of subject-predicator concord is breached in order to describe
the referent of a plural noun group as a unit rather than as a plurality. This
type of notional concord is found not only with plural nouns referring to time
but also to other types of quantity or measure:
(12) Three square miles of marsh was bought in 1995.
(13) Two pints of milk has always been sufficient.
If the subject group begins with a singular determiner, it is nearly always the
-s form of the verb which is selected, and if the sentence begins with a
provisional subject realized by there, there is usually singular concord too
(see Juul and S0rensen 1978: lOOff):
(14) Another fifteen pounds is due tomorrow.
(15) There's two pints of milk in the fridge.
While plural expressions of quantity or measure very often select the -s form,
the speaker may also prefer to focus on the separate members of the set
referred to and thus observe the rule of syntactic concord:
(16) The first six months were spent in India.
(17) Two pints of milk have always been sufficient.
C) Plural proper nouns. As pointed out in section 7.6.3, geographical
proper nouns in the plural form like the Andes, the Alps and the Hebrides
select the -0 form. With some geographical plural nouns, however, the
speaker has a choice between the -0 form and the -s form and can in this way
focus on a set as a unit or on the separate members of a set. To these nouns
belong the Netherlands, the Midlands and Kew Gardens. The combination of
such nouns with the -s form and the -0 form can be illustrated by the
following examples (cf. Juul 1975: 260):
(18) The Midlands reflects the same picture of poverty and misery.
(19) The Midlands attract surprisingly many tourists.
The United States nearly always selects the -s form of the verb (except where
one of its national sports teams is referred to) and so does the United
Nations.
Titles of books, plays, etc. select the -s form of the verb if they are
regarded as names, i.e. are used as proper nouns: 'The Three Musketeers' is
undoubtedly Dumas' best known work / 'Ghosts' was produced last night at
the National Theatre / My Canterbury Tales is on the table. But if they are
used to describe a literary production as consisting of a number of separate
parts, it is usually the -0 form of the verb which is selected: The Canterbury
Tales contain several bawdy stories. Names of companies, institutions, etc.
in the plural form behave like collective nouns with respect to subject-
predicator concord: British Airways expect(s) still more customers next year.
D) Plural 'fact1 expressions. We also find notional concord in:
(20) Many cars on the roads means many traffic accidents.
If this sentence is compared with Many cars on the roads are in bad repair,
it appears that something seems to be missing from the subject realization,
for example The presence of... . Behind the plural expression there appears
to lie a singular concept which explains the selection of the -s form of the
verb. Reference is made to a fact or circumstance, and the meaning of the
plural subject expression can therefore be captured by the paraphrase 'The
fact that there is/are x', i.e. by a noun group in which the head is realized by
the singular form of FACT. Another possible paraphrase is 'That there is/are
x', and as this is a clause, the selection of the -s form makes sense as well.
Plural 'fact' expressions are particularly common in sentences where the
predicator is realized by MEAN (or related verbs like ENTAIL, IMPLY,
INVOLVE), but we find it in sentences with other verbs as well:
(21) High production costs prevents reasonable consumer prices.
E) Compound units with and. As mentioned in section 7.6.3 [B], a
coordinated construction with and in which the conjoints are realized by
singular nominals selects the -0 form of the verb. In those cases where the
referents of such a construction are regarded as constituting a unit, however,
notional concord overrules syntactic concord:
(22) Bed and breakfast is provided at fifteen pounds.
Notional concord of this type is found with coordinated noun groups

referring to meals (bacon and eggs, fish and chips), drinks (gin and tonic,
whisky and soda, rum and cola), pubs (the Spade and Bechet, the Fox and
Goose) and with other established expressions such as board and lodging,
trial and error and the Stars and Stripes. Names of companies, etc. (e.g.
Harland and Wolff), behave like collective nouns.
If a compound unit with and is used to refer to one person or thing only, it
is naturally enough the -s form of the verb which is selected:
(23) My colleague and friend, Ian Mackay, has just published another book.
7.6.5. Attraction
In section 7.6.3 [B], we noted attraction of verb form to the last conjoint in
certain compound units (e.g. You or your partner is crazy instead of You or
your partner are crazy). Similarly, in section 7.6.3 [D], we mentioned that
pronouns like no one, everybody, either, neither, etc. are more likely to select
the -0 form if they are postmodified by a preposition group in which the
dependent contains a plural noun than if they realize the subject on their own
(recall the examples Neither seems qualified for the job and Neither of the
applicants seem qualified). We speak of attraction when a form other than
the one we would normally expect to determine concord exerts decisive
influence on the form of the verb. Typically, in the case of subject groups,
such a distracting form intervenes between the head of the subject and the
predicator, and the longer the distance between the subject head and the
predicator, the more likely it is that the speaker will let this form determine
the form of the verb. In examples like the ones with pronouns already given,
attraction can hardly be considered incorrect. But in an example like the
following — which is characteristic of unplanned discourse and not
uncommon - it is generally considered incorrect:
(1) The situation in Bosnian mountain areas and forests now seem critical.
Attraction to a post-head dependent may also be exerted by a compound unit
with and in which the conjoints are realized by singular nominals:
(2) The systematic study of grammar, phonetics, semantics and linguistics are
generally considered indispensable to university students of a foreign
language.
It may also be a singular form which determines the choice of verb form in
sentences where the subject is realized by a plural form:
(3) He is one of those students who never prepares for class.
In the relative clause of this complex sentence the subject is realized by who,
the antecedent of which is the plural form students and not one of those
students, the head of which is the singular pronoun one. Though very
frequent, this type of attraction is often considered incorrect too. Note that it
is here a more distant word which 'attracts' the verb with respect to number.
In order to appreciate the difference between a one-word antecedent immedi-
ately preceding the relative pronoun and one whose head word is separated
from the pronoun by a postmodifying construction, it is useful to compare
noun groups like:
(4) clauses beginning with a wA-word which functions as object
(5) clauses beginning with a wA-word which have the structure S P C
A third type of attraction is found in sentences in which a subject
complement is realized by a plural form (see Juul and S0rensen (1978: 95),
who cite the following example):
(6) Markoffs material were 20,000 letters comprised in the first chapter and the
first 16 sonnets of the second chapter...
In S P C sentences it is regularly the subject which determines the choice of
verb form (though the fossilized biblical expression The wages of sin is death
may be noted). Attraction of the type illustrated by (6) is generally
considered incorrect and is therefore 'fatal' too, like that illustrated by
examples (1), (2) and (3).
7.7. Other types of external concord

In those cases where a subject complement is realized by a noun or noun
group there is usually subject - subject complement concord of number:
(1) However, the latest setback is a further embarrassment to the troubled
economy.
(2) The banana plantation is a reservoir of rats.
(3) The liaison may even prove a good career movement.
(4) The warehouses and storerooms are perfect breeding grounds for rats.
(5) The children have become good friends.
(6) They proved very amusing companions.
As in the case of subject-predicator concord, however, there are many
exceptions to this syntactic rule. Here as well the semantic nature of the noun
groups involved plays an important role (see Huddleston 1984: 187). An
example like Laurel and Hardy are an amusing pair is perfectly natural, for
'an amusing pair' can obviously only be predicated of two entities, and Laurel
and Hardy constitute a unit (though notional concord does not here, as in the
Other types of external concord 251
case of bed and breakfast, etc., overrule syntactic subject-predicator

concord). In Laurel and Hardy are a good choice the complement is realized
by a noun group which is usually predicated of a single entity, so in this
example discord of number can only be explained semantically as being due
to the unitary conception of Laurel and Hardy. Altogether S P C sentences in
which S is realized by a plural form and C by a singular form are not
uncommon. While examples (5) and (6) could not be changed, the subject
complement of example (4) could also have been realized by a singular noun
group: The warehouses and storerooms are a perfect breeding ground for
rats. Note also that while e.g. (Both) Woody Allen and Clint Eastwood are
good choices - where the referents of the subject nouns can hardly be
regarded as constituting a unit - is a more natural sentence than (Both)
Woody Allen and Clint Eastwood are a good choice, the latter cannot be
ruled out as ungrammatical (Huddleston 1984: 187).
While sentences in which S is realized by a plural form and C by a singular
form are thus fairly common, the opposite kind of discord is practically never
found. Apart from idiomatic expressions like Thatcher is friends with
Reagan it only occurs in examples of the type illustrated by a sentence like
The policy I advocate is lower tariffs.
In S P O C sentences there is usually object - object complement concord
of number:
(7) God has made me a cripple.
(8) God has made them cripples.
Here as well, however, we find syntactic discord - in the shape of Ο realized
by a plural and Co realized by a singular form:
(9) This makes Laurel and Hardy a good choice.
Unlike a complement form, an object form selects its number independently
ofthat of the subject form, as illustrated by examples like That guy has killed
my best friends and Those guys have killed my best friend. Only in the case
of the so-called distributive plural is there concord of number between S and
Ο (for more details on the distributive plural, see section 10.4.6):
(10) The terrorists lost their lives.
(11) Such clauses rarely have third person subjects.
In these sentences - which should be compared with The terrorist lost his life
and Such a clause rarely has a third person subject - the members of a set of
entities denoted by one constituent form are matched one by one by those
denoted by another form. This type of plural concord is found not only in S P
Ο sentences but also where other constituents are involved. Examples
illustrating this are We complained about our wives and The students were
shouting at the top of their voices (S P A). In some cases it is possible either
to focus on the one-by-one correspondence between the members of two sets
or to refrain from doing so: The monks bowed their head(s) in shame. Where
idiomatic expressions are involved, the rule of the distributive plural tends to
be ignored: The inspectors can 'tput their finger on the cause of trouble / The
inspectors put their foot down.
In S P O sentences like She shot herself 'and They shot themselves there is
agreement of number, person and gender between the reflexive pronoun
realizing the object and the subject form. This is not an instance of subject-
object concord, however, but due to a more general rule that a pronoun
agrees with its antecedent. This rule also explains the agreement found in
examples like She is not herself these days, They gave themselves a break,
I'll pay for myself and We finished the job ourselves where the reflexive
pronouns realize a complement, an indirect object, a prepositional com-
plementation and part of a subject respectively.
8. The complex sentence
8.1. Introduction
By a complex sentence we understand a sentence with one or more sub-
clauses. There are two types of complex sentence: that in which one or more
sentence functions have clausal realization (e.g. Orel) and that in which this
is not the case but where one or more group constituents are realized
clausally. The latter type of complex sentence can be illustrated by:
(1) The residents living in the poorhouse avoided her like a leper.
(2) As soon as he gave me that guitar I forgot about piano.
(3) The passengers were marshalled into what appeared to be the arrivals hall.
(4) His giving me this note is an act of defiance.
In the first three examples, the italicized clause realizes the dependent of a
group: a noun group in (1), an adverb group in (2) and a preposition group in
(3). In example (4), the italicized clause realizes the head of a group
determined by the dependent pronoun His. Complex sentences of the type
exemplified by (1) through (4) will be dealt with in chapters 10 through 12
on word classes and groups.
In this chapter we examine complex sentences of the first type, i.e. sen-
tences in which one or more sentence functions have clausal realization. We
note, first of all, that the same seven basic patterns are found in complex
sentences as in simple sentences:
SP To see her falling in love hurts.
SPA Going abroad with her was on the agenda.
S PC The question is whether she wants him back.
S PO We must assume that elsewhere all hell is being let loose.
S POO He gives whoever turns up after office hours a rough ride.
S POA Lucy spent that summer writing letters of application.
S POC We'll elect you whatever you like.
As we see in these examples, any of the sentence functions S, Od, Oi, Cs, Co
and A (but not P) may have clausal realization.
We also very frequently find clausal realization of optional adverbials:
(5) As dusk fell, the group's mood became more querulous.
(6) To be quite honest, I hope they think we're travelling together.
Some complex sentences can be found which do not readily fit into one of
the seven basic patterns. A case in point is sentences in which an indirect
254 The complex sentence
directive is expressed by verbs like ADVISE, ASK, BEG, ENTREAT, PER-

SUADE, REMIND, REQUEST, TELL, URGE followed by a noun group + a ίο-
infinitive clause (see Quirk et al. 1985: 1215):
(7) They persuaded the young man to see a psychiatrist.
It is not entirely obvious how such sentences should be analysed. An analysis
in terms of an S P O C structure seems excluded because there is no implied
S P C construction in the structure following the finite verb, as there is in an
ordinary S P O C sentence like They made the young man their leader (see
section 3.2.8). But should it then be classified as an S P O sentence or as an S
P O O sentence? If an S P O analysis is adopted, the problem arises that the
entity affected by the action expressed by Ρ is not everything following the
finite verb, as it is in a S P O sentence like They wanted the young man to see
a psychiatrist, but the young man exclusively (see section 3.2.5 on the
identification of objects). Note also that the infinitive clause can be omitted
without changing the semantic relation between persuaded and the young
man. If (7) is analysed as an S P Oi Od sentence, we should expect the young
man to be a BENEFICIARY DONE-TO like the indirect object in an
ordinary S P O O sentence like They gave the young man what money they
had. Semantically, the role performed by the young man in (7) is instead
simultaneously AFFECTED DONE-TO and AGENT DOER. The problem
of classifying examples like (7) arises because two clausal structures overlap:
cl1
They persuaded | the young man to see a psychiatrist ,
cl2
Rather than assign examples of this type to one specific pattern, we suggest
that they belong in a grey zone between the S P O and the S P O O patterns.
8.2. Classification of subclauses

Formally, subclauses may be divided into finite clauses, non nite clauses
and verbless clauses. The most common types of finite clauses are that-
clauses, independent relative clauses and interrogative clauses:
(1) She discovered that they were listening with apparent interest.
(2) What you said kept me awake all night.
(3) Whether your plan will succeed remains to be seen.
(4) I've discovered who my real friends are.
As illustrated by (3) and (4), there are both yes-no interrogative subclauses
(signalled by if ana whether) and wA-interrogative subclauses (signalled by
Classification ofsubclauses 255
interrogative pronouns and WHEN, WHY, HOW). The subordinating conjunc-

tion is in some contexts omitted from thai-clauses, particularly if the clause it
introduces realizes an object (see also section 6.4.1).
Nonfinite clauses may be divided into -ing participle clauses (also referred
to as present participle clauses), -ed participle clauses (or past participle
clauses), to-infinitive clauses and bare infinitive clauses:
(5) Going home that evening, I stopped at the chemist's for some razor blades.
(6) The soldiers were impassive unless challenged by the passengers.
(7) I was only trying to forestall alarm and despondency.
(8) All I did was ask them round for drinks.
As illustrated by these examples, nonfinite clauses are usually subjectless,
but all four types permit overt subjects:
(9) Our chairman being away on holiday, there is nothing we can do right now.
(10) Dinner finished, the fellows adjourned to the combination room.
(11) I want you to leave immediately.
(12) Rather than he call her, I prefer to tackle this problem myself.
As will appear below, -ing participle clauses and ίο-infinitive clauses are
commoner than -ed participle clauses and particularly bare infinitive clauses.
Verbless clauses are nearly always adverbial and may or may not contain a
subject form:
(13) Ronald knelt down, Ais hands behind his back.
(14) Although invariably very considerate, she strikes me as somewhat cold.
Note also the existence of examples like the following in which a verbless
clause with adverbial properties could be viewed as a reduced 'nonrestrictive
relative clause1, i.e. a clause which provides further information about the
person or entity it refers back to (cf. sections 10.1.4 and 11.3.3 [A]):
(15 a) George, ready for action now, stared at the intruder.
(15b) George, who was ready for action now, stared at the intruder.
According to syntactic function, subclauses realizing sentence functions can
be subdivided into subject clauses, direct object clauses, indirect object
clauses, subject complement clauses, object complement clauses and
adverbial clauses. The first five types are traditionally referred to as 'nominal
clauses', because the sentence functions involved are typically realized by
nominals. In our discussion below we select the functional classification of
subclauses into these six types as our point of departure.
8.3. Subject clauses

Subjects realized by nite clauses can be illustrated by:
(1) That you dislike her is obvious to everybody in the department, (thai-clause)
(2) What you said kept me awake all night, (independent relative clause)
(3) When you can help me is what I want to know. (wA-interrogative clause)
(4) Whether she will recover is still an open question, (yes-no interrogative
clause)
In some cases it is difficult to tell whether a subject clause is realized by an
independent relative clause or by a wA-interrogative clause (see Quirk et al.
1985: 1059ff). This is particularly the case if the w>A-word involved is what.
For example, a sentence like What he said was unclear is ambiguous in
written English in that it can mean either 'That which he said was lacking in
clarity' or Ί didn't get what he was trying to say'. In spoken English, we
should add, this example is unambiguous, for here what is unstressed if the
subject clause is relative and is pronounced with nuclear stress if the subject
clause is interrogative. Syntactically, independent relative clauses realizing a
subject differ from interrogative wA-clauses performing this sentence
function in that they select the plural form of the verb if a noun determined
by what is in the plural, thus behaving more like groups than clauses (see
section 7.6.3 [C]):
(5) What enemies he has have left the department.
(6) What enemies he has is hard to tell.
Furthermore, independent relative clauses in which what determines a noun
differ from interrogative w/mf-clauses of this type in that they may be
followed by few and little. For example, few can be inserted after what in (5)
but not in (6), and little can be inserted after what in What money he has will
soon be spent but not in What money he has is hard to tell. Semantically,
interrogative wA-clauses differ from independent relative clauses in that they
contain a gap of unknown information (Quirk et al. 1985: 1060).
As in the case of subject clauses, it is sometimes difficult to tell whether an
object clause is realized by an independent relative clause or a wh-
interrogative clause. For example, a sentence like He asked me what I
expected (quoted from van Ek & Robat 1984: 47) is ambiguous - not only in
written but also in spoken English - in that it can mean either 'He asked me
that which I expected* or 'He asked me "What do you expect?'".
In those cases where a subject is realized by a nonfinite clause, the
subtypes of such clauses are -ing participle clauses and ίο-infinitive clauses:
(7) Getting on with things is just what we can't do.
Subject clauses 257
(8) To say that my heart isn 't in it would be the understatement of all time.
Continuous to-infinitive clauses with subject function do not have subjects.
Arguably, discontinuous to-infinitive clauses have subjects in cases like the
following (for discussion, see section 5.6.1):
(9) John happened to be around at the time.
(10) Sally is believed to be guilty.
In continuous to-infinitive clauses, an AGENT (or another participant with
subject-potential in a corresponding finite clause) may be expressed by an
initial adverbial group with the preposition/jr as head, compare:
(11 a) To step down now would be frowned upon by everybody.
(lib) For Rita/her to step down now would be frowned upon by everybody.
That For is here a preposition rather than a conjunction appears from the fact
that it comes and goes with the following (pro)nominal constituent and from
the fact that the infinitive clause is already marked as subordinate by virtue
of being nonfinite and therefore has no need for a semantically empty
subordinator conjunction. Like the very similar by-group in passive con-
structions, the^frr-group is syntactically optional.
The next example illustrates subject realization by an -ing clause which has
its own subject:
(12) John/Him being Jewish makes no difference, (cf. Vestergaard 1985: 216)
This should be compared with John's/His being Jewish makes no difference,
where the subject is not realized by a clause but by a group in which the
dependent is realized by a genitive or possessive pronoun determiner
(John 's/His) and the head by the -ing participle clause being Jewish (cf.
John 's/his Jewishness makes no difference). As regards the use of these
competing constructions in subject clauses, it should be mentioned that (i)
possessive pronouns are usually preferred to personal pronouns, (ii) the
genitive form is preferred to the uninflected nominal form, especially in
formal style, unless the nominal has impersonal or inanimate reference (see
section 10.3.8 on the genitive). It should be added that while the genitive is
thus generally preferred to the uninflected form of a nominal in subject
clauses, the opposite is the case in clauses realizing other sentence functions:
(13) John 's being Jewish makes no difference.
(14) I remember John telling me that joke.
As pointed out in section 5.2.3, a subject is often moved from its neutral
position to the end of the sentence for reasons of end-weight and/or end-
focus. Such extraposition can be illustrated by the following examples in
which the postponed subjects are realized by finite clauses:
(15) It has been known that there are groups out there that are willing to kill at
random.
(16) It doesn't matter who fired him.
(17) It's uncertain when we can tackle this problem.
(18) It's still an open question whether she will survive.
With ίΑαί-clauses, extraposition is the norm rather than the exception and
may even be obligatory, for example in It seems/appears that the house was
set on fire. Interrogative clauses of both types (yes-no; wh-) are also freely
postponed, as illustrated by examples (16), (17) and (18), but extraposition of
independent relative clauses is not common. If a sentence like It was unclear
what he said is pronounced neutrally with nuclear stress on said, or if it is
pronounced with contrastive nuclear stress on what, the subject clause will
be understood to be interrogative (Ί didn't get what he was trying to say').
Only if the entire subject clause is pronounced without prosodic prominence
and the nuclear stress falls on (un)clear, can it be understood to be an
independent relative clause ('That which he said was lacking in clarity').
Extraposed subject clauses may be nonfmite:
(19) It frightens me to think how vulnerable we are.
(20) It's splendid seeing you back on form.
While ίο-infinitive clauses are freely postponed, extraposition of -ing
participle clauses is much rarer and is usually characteristic of informal style
(van Ek & Robat 1984: 412). In e.g. Getting on with things is just what we
can't do, postponement of the subject clause would have the effect of
demoting the part of the message expressed by this clause to an explanatory
appendage: It's just what we can't do, getting on with things. Such right-
dislocation (cf. section 4.5) is not motivated by weight as it is in the case of
extraposed clauses.
8.4. Cleft sentences

In this section we shall examine a special type of subject clause. Consider
first a complex S P C sentence like the following in which the subject is
realized by an independent relative clause beginning with what:
(1) What worries me is the poor quality of your work.
If this is compared with the simple sentence The increasingly poor quality of
your work worries me, we notice an important difference in information
structure: the part of the message realizing the noun group is highlighted
through final position (end-focus). Highlighting (combined with identi-
Cleft sentences 259
fication) can also be obtained in another way, however, namely by means of

a so-called cleft sentence:
(2) It is the poor quality of your work that worries me.
A cleft sentence is a sentence in which a constituent singled out for emphatic
identification is placed between it + BE and a subclause realizing the
remainder of the corresponding simple sentence. It, it should be added, has
no reference outside the sentence, and in this way it differs from it used in
superficially similar but noncleft sentences. In a noncleft sentence like It is
Mr Blake, who has just been appointed - used in reply to e.g. Who is your
boss? - It refers to your boss in the preceding question (see N01ke 1984: 73).
Sentences like (2) are called 'cleft' because they are divided into two parts.
Typically, the second of these, the subclause, expresses given information
and is pronounced without prosodic prominence, whereas the focalized
element in the matrix clause (such as the poor quality of your work in (2))
expresses new information and is pronounced with nuclear stress.
By means of the cleft construction not only subjects but also objects and
adverbials may be focalized, cf. the following cleft versions of Russell met
Keynes in Cambridge:
(3) It was Russell that/who met Keynes in Cambridge.
(4) It was Keynes (that) Russell met in Cambridge.
(5) It was in Cambridge (that) Russell met Keynes.
Prepositional complements may also be focalized, as shown by an example
like It is her last novel (that) I'm interested in. Indirect objects, subject
complements and object complements are rarely focalized by cleaving in
English, but examples like It's me you do a great honour and It's president of
the board of directors he's become/they've elected him can hardly be
considered unacceptable. Predicators cannot be focalized by cleaving at all.
Cleft sentences are not declarative exclusively but may also be inter-
rogative:
(6) Is it the quality of his work that worries you?
Furthermore, cleft constructions may be subordinated:
(7) I hope it isn't the quality of my work that worries you.
Cleft sentences are nearly always used for the same purpose, namely to
signal an identification. In (3), for example, the person who met Keynes in
Cambridge is identified as Russell. Another property shared by cleft
sentences is that in addition to describing a situation they contain a
presupposition, i.e. a proposition regarded as true by the speaker, which is
both positive and negative (Vikner 1973 and Hansen, ms.). By uttering (3),
for example, the speaker assumes not only that someone met Keynes in
Cambridge but also that there was somebody else who did not do so. Any
positive cleft sentence is thus matched by a negative cleft sentence.
Conversely, any negative cleft sentence, for example // wasn't me who did it,
is matched by a positive cleft sentence.
Sometimes the subclause is omitted from a cleft sentence. Such a reduced
cleft is context-bound in that it can only be used if the subclause can be
recovered from the context (see Nelke 1984: 74). For example, It wasn't me
can be used instead of an unreduced cleft sentence in reply to a question like
Who broke the window?
As a cleft sentence expresses an identification, it is obvious that the form
focalized should permit such identification, i.e. it should refer to a person, a
thing, a place, a cause, etc. (see Vikner 1973). A number of restrictions on
the cleft construction can be explained in this way. For example, disjuncts
like possibly, fortunately, in all likelihood and conjuncts like nevertheless,
though, anyhow do not permit such referential identification and con-
sequently cannot be focalized in clefts. If a subject or an object is realized by
a clause (which describes a situation), cleaving is heavily restricted. For
example, we cannot by means of cleaving focalize the object clause of a
sentence like He tried to concentrate upon rational assessment of the
situation. On the other hand, clausally realized adjuncts are frequently
focalized by means of the cleft construction:
(8) It is because you drink so much that I avoid your company.
(9) It is only when you drink that I avoid you.
A number of restrictions on clefts can be explained semantically (see
Hansen, ms.). For example, negative expressions like nobody, nothing,
nowhere, in no way and never cannot be focalized as this would lead to a
contradiction of the positive presupposition contained in the subclause. Thus
*It is nobody that/who loves me is not a possible cleft version of Nobody
loves me because it contradicts the presupposition that somebody loves me.
Similarly, indefinite expressions like somebody, somewhere, anyone and
anywhere are incapable of being focalized in clefts, as illustrated by the
unacceptability of e.g. *It is somebody who loves me (cf. Somebody loves
me). In this case the constraint is due to the fact that somebody, somewhere,
etc. do not provide any identification because no option is eliminated. Note,
however, that while e.g. *It is somewhere that you will find the gun is
unacceptable, this is not the case with It is somewhere in here that you will
find the gun. The reason for this is that somewhere is postmodified
restrictively by a preposition group which provides the necessary identi-
fication and contrasts with e.g. out there.
Cleft sentences 261
Let us turn now to the syntactic properties of cleft sentences. The subclause
is usually analysed as a restrictive relative clause (cf. section 11.3.3 [A]), but
it differs from regular clauses of this type in three ways:
(i) It may be added to a proper noun (e.g. It is John that has resigned).
(ii) There is often a preference for that or zero instead of who and which,
except if it serves as subject.
(iii) It may be added to an adverbial clause, as in It is only when you drink
that I avoid you (see Quirk et al. 1985: 1387).
In those cases where an adjunct is focalized, it is perhaps not immediately
obvious whether that should be analysed as a relative or as a subordinating
conjunction. On reflection, however, it seems clear that it has its own
syntactic function inside the subclause and that this function is the same as
that of the focalized element. In a sentence like It was in Toronto that I met
her, that is thus a relative, and its function in the subclause is adverbial, like
that of in Toronto in the matrix clause. If that is omitted, this analysis also
applies to zero: // was in Toronto 01 met her.
Having analysed the subclause of a cleft sentence as a special type of
restrictive relative clause we must consider what its antecedent is and - more
generally - how cleft sentences should be analysed syntactically. The
analysis which comes to mind first is the straightforward one that the
highlighted element is the antecedent and the subclause is a post-head
dependent in a group realizing a subject complement. Like other relative
clauses, the relative clause in a cleft construction shows concord relations
with its antecedent:
(10) It is my brother who wants to go on a holiday.
(11) It is these books which challenge his views.
If this analysis is adopted, however, we run into serious difficulty. This can
be illustrated by an example like It was the vase that Agatha gave us (quoted
from Huddleston 1984: 461). In written English this is ambiguous, for it may
represent not only a cleft sentence but also an ordinary uncleft sentence in
which the subclause restricts the reference of the preceding noun ('It was the
vase given to us by Agatha1). We would not want the same syntactic analysis
for both. In the cleft version the subclause is more loosely connected with the
vase than in the noncleft version, as shown by the fact that the focalized
element can be fronted: The vase it was that Agatha gave us. This seems to
indicate that the subclause is not part of the same constituent as the focalized
element. In other words, the problem at hand is how to show that the
focalized element and the relative clause are at the same time closely related
and yet two separate constituents. One possibility is to employ stacking:
STA
It
vase that Agatha gave us
The closeness relation between the focal element and the relative clause is
here reflected by means of a form and function stack. The separate status of
the two constituents is then acknowledged at a lower level by assigning
different functions to them within the stack. The focal element, which
expresses new information and follows a form of BE, is assigned subject
complement status. The relative clause, which expresses given information,
is interpreted as a real subject in fixed position. The pronoun It is simply a
form word representing the information expressed by the relative clause (It =
'what Agatha gave us'). Note that this information could have been expressed
by an independent relative clause in normal subject position (e.g. What
Agatha gave us was the vase), but with unmarked end-focus on the vase.
In closing this section we return to examples with independent relative
clauses like the one just mentioned, as well as (1) above: What worries me is
the increasingly poor quality of your work. As these examples are closely
related to the simple noncleft sentences Agatha gave us the vase and The
poor quality of your work worries me, not only semantically but also
syntactically, they would seem to represent a second type of cleft sentence.
In many cases, however, sentences of this type are unmatched by simple
sentences, as illustrated by an example like What went wrong was that the
valve was overheated (cf. Huddleston 1984: 464). Note also the functional
diversity of the independent relative clause: in the examples discussed so far
it serves as subject, but often it might equally well serve as subject
complement instead, in which case it receives unmarked end-focus, e.g. The
poor quality of your work is what worries me. For these reasons we do not
classify them as cleft. Such sentences - traditionally termed pseudo-cleft -
Object clauses 263
may be regarded as a special case of sentences whose subject is realized by

an independent relative clause with what.
Pseudo-cleft sentences differ from cleft sentences in that they make it
possible for the entire predicate to be singled out for identification:
(12) What she did was (to) tell me off in public.
If this sentence is compared with the simple sentence She told me off in
public, we see that it is the entire predicate stack which is focalized by means
of the finite proform of DO in the relative clause. Another characteristic of
pseudo-cleft sentences is that the subject complement may be realized by a
clause. Such a clause may be nonfinite, as in (12), or it may be finite:
(13) What bothers us is that you take no interest in your work.
In truly cleft sentences a subject complement cannot be realized clausally, cf.
the ungrammaticality of *It is that you take no interest in your work that
bothers us.
8.5. Object clauses

Object clauses realized by finite object clauses can be illustrated by the
following examples:
(1) Police believe (that) the shooter was a professional, (that-clause)
(2) The child does what society says he must, (independent relative clause)
(3) Keeping a nervous eye on passing traffic, many wondered out loud which of
them would be next, (wit-interrogative clause)
(4) I'll ask her if she'll lend us the money, (yes-no interrogative clause)
Objects may be realized by all four subtypes of nonfinite clauses, but to-
infmitive clauses are particularly frequent:
(5) I enjoy travelling by train.
(6) I want this reported to the police.
(7) We watched the sun set behind the mountains.
(8) I don't want to become a refugee for the rest of my life.
Continuous to-infinitive object clauses may have their own subject (unlike
corresponding subject clauses, cf. section 8.3):
(9) I don't want you to become a refugee for the rest of your life.
Recall that a clausally realized object may be postponed to secure end-weight
and that in those cases where there is a provisional object in the shape of it,
object postponement involves extraposition (see sections 3.2.5 and 5.2.3):
(10) I just saw on television how some Indian people started a shop and put the
old grocery on the corner out of business. (S A P A O)
(11) I find it a challenge writing this report. (S P Op Co Or)
Verbs vary considerably with respect to what forms of clausal com-
plementation they take. For example, AVOID requires an -ing clause and
ANSWER a fAaf-clause:
(12) I avoided drinking wine.
(*I avoided to drink wine / *I avoided that I drank wine)
(13) She answered that she joined the party.
(*She answered to join the party / *She answered joining the party)
WANT may take an object ίο-infinitive clause, in some cases a participle
clause, but not a f/w/-clause:
(14a) I want you to leave at once.
(*I want that you leave at once)
(14b) I don't want you arriving late.
(14c) I want it done now.
Other verbs that allow of more than one type of clausal complementation:
(15a) I believe him (to be) guilty of murder.
(15b) I believe that he is guilty of murder.
(16a) She liked to swim in the morning.
(16b) She liked swimming in the morning.
In these examples there is very little semantic difference between the options.
In other cases, there is a clear difference of meaning:
(17a) I remembered to post the letter. (= Ί did not neglect to post it')
(17b) I remembered posting the letter. (= Ί looked back on the event')
(18a) She tried to close the window (but it was stuck).
(18b) She tried closing the window (but there was still too much noise coming
from the street).
Note that different types of clausal complementation sometimes assume dif-
ferent sentence functions, resulting in an even clearer difference of meaning:
(19a) Jack stopped to examine the results. (S P A)
(19b) Jack stopped examining the results. (S P O)
(20a) Sally went on to discuss the children. (S P A)
(20b) Sally went on discussing the children. (S P O)
Finally, it should be mentioned that objects are realized clausally in (directly
or indirectly) reported speech:
(21) 'I'm no puritan,' she answered.
Subject complement clauses 265
(22) She says (that) she would like another whisky.

For discussion, see sections 5.3.9. and 5.6.1.
8.6. Subject complement clauses

Subject complements can be realized by most clause types. The following
examples illustrate realization by finite clauses:
(1) The point is that we are now involved. (fAaf-clause)
(2) All I can think of is what you said last night, (independent relative clause)
(3) My worry is how you could do a thing like that. (wA-interrogative clause)
(4) The question is whether she will recover, (yes-no interrogative clause)
In subject complement ίΑαί-clauses the subordinating conjunction cannot
readily be omitted as it can if such a clause realizes an object.
Subject complements may also be realized by -ing participle clauses, to-
infmitive clauses and - in very specific contexts - bare infinitive clauses:
(5) The dividing line between a good and a bad politician is knowing what can
be achieved with the means at his disposal at any given time.
(6) My only aim in life is to free my children.
(7) What she did was (to) tell me off in public.
(8) All I did was (to) ask them round for drinks.
The realization of a subject complement by a bare infinitive clause typically
requires the presence of a form of DO in the realization of the subject, either
in an independent relative clause beginning with what (as in (7)) or in a
postmodifying relative clause (as in (8)). In such examples, the subject
complement may also be realized by a to-infinitive clause.
Present participle clauses realizing a subject complement may have an
overt subject:
(9) A crucial event was Gascoigne being sent off for a professional foul.
To-infinitive clauses realizing a subject complement do not have subjects
(the same way that clauses of this type with subject function do not have
subjects, see section 8.3). In an example like the next one, for Rita is a
preposition group realizing an adverbial within the complement clause:
(10) The only solution is^br Rita to step down.
As appears, a complement to-infinitive clause may express an agent (or some
other participant with subject potential) in an initial adverbial far-group.
8.7. Indirect object clauses and object complement clauses

In the case of indirect objects and object complements the possibilities of
clausal realization are heavily restricted. Both can be realized clausally by an
independent relative clause or by an -ing participle clause:
(1) He gives whoever turns up after office hours a rough ride.
(2) She gave going to France a good deal of thought.
(3) You can paint the wall whatever) colour you like.
(4) I would call that casting pearls before swine.
Arguably, an object complement can be realized by an -ed participle clause:
(5) I keep the will hidden in my drawer.
The analysis of this sentence poses a number of problems, but it seems most
natural to assume that the will realizes an object and hidden in my drawer an
object complement. Note that the semantic relationship between these two
word sequences is of the same nature as that between the will and safe in an
S P O C sentence like / keep the will safe. As hidden must be analysed as a
passive participle rather than as an adjectival -ed form (see section 7.4.6), the
object complement in (5) is realized by a clause in which hidden functions as
predicator and in my drawer as adverbial.
8.8. Adverbial clauses

Adverbial clauses are extremely frequent and may be finite, nonfmite or
verbless. Finite adverbial clauses are typically introduced by subordinating
conjunctions (cf. section 6.4.1):
(1) A glider pilot is in hospital with a fractured rib after his aircraft crashed into
an electricity pylon.
(2) I left my wife because I realized that I had made an awful mistake.
An adverbial may be realized by the following subtypes of nonfmite clause:
(3) Keeping a nervous eye on passing traffic, many wondered out loud which of
them would be next, (-ing participle clause)
(4) The soldiers lounging at the entrance were impassive unless challenged by
the passengers, (-ed participle clause)
(5) The family enrolled in courses to attain enlightenment, (to-infinitive clause)
If rather than is interpreted as a complex conjunction (and not as an adverb
followed by conjunction) on a par with instead of, the adverbial in the
following example is realized by a bare infinitive clause:
(6) Rather than call her, you should invite her over for lunch.
Adverbial clauses 267
Realization of an adverbial by a verbless clause can be illustrated by:

(7) Although invariably very considerate, she strikes me as somewhat cold.
Nonfinite and verbless clauses may also have their own subject (see section
8.2), typically in formal style:
(8) Our chairman being away on holiday, there is nothing we can do right now.
(9) Dinner finished, the fellows adjourned to the combination room.
(10) Ronald knelt down, his hands behind his back.
A nonfmite or verbless adverbial clause which has its own subject and is not
introduced by a subordinating conjunction is traditionally called an absolute
clause. In this connection attention should also be drawn to so-called
supplementive clauses, i.e. participial and verbless clauses without a sub-
ordinator like the ones occurring in examples such as This passage is
unclear, taken out of context and / slammed down the receiver, furious after
a day of frustrations (see Quirk et al. 1985: 1123ff). Such clauses - which
may also have their own subject and thus be absolute - describe attendant
circumstances the value of which (conditional, temporal, causal, etc.) has to
be inferred from the context.
In nonfinite or verbless adverbial clauses without an overt subject, the
'understood' subject form normally has the same reference as the subject
form of the superordinate clause, as in (3) through (7). As illustrated by the
next examples, however, this is not always so:
(11) Keeping a nervous eye on passing traffic, it suddenly occurred to me that my
own family might be next.
(12) Known primarily as the author of 'Changing Places' and 'Small World',
many consider Lodge a humourist and writer of campus novels.
(13) Taking his recent record into account, Major appears to be the wrong man
for the job.
In cases like these it is customary to speak of unattached participles (or
'dangling participles'). This term is not entirely appropriate, however. In (11),
the hearer is able to attach the -ing participle clause to the words me and my
in the matrix clause and infer that the understood subject of the subclause is /
(As I was keeping a nervous eye on passing traffic, it...). In (12), similarly,
there is something to hang on to in the matrix clause, namely the object noun
Lodge, so here as well the hearer is capable of inferring the understood
subject of the subclause. In (13) there is no noun or pronoun in the matrix
clause on the basis of which the subject form of the subclause can be
inferred; but here the hearer will correctly assume that it is a pronoun with
generic reference (one, you or we).
While (13) is unproblematical, and while (11) and (12) are acceptable to
most speakers of English, an example like the following is unacceptable, or
at least highly questionable:
(14) Known primarily as the author of 'Changing Places' and 'Small World', my
brother considers Lodge a humourist and writer of campus novels.
What is wrong here is that the understood subject of the participle clause
may by the hearer or reader be understood to have the same reference, not as
the object Lodge in the superordinate clause but as the subject my brother. In
this way (14) differs from (12), in which there is no risk that the understood
subject of the participle clause may be assumed to have the same reference as
the subject many in the superordinate clause.
An unattached participle is also found in an example like the next one
(quoted from McArthur 1992: 753):
(15) There, coasting comfortably down the attractive green coastline, the town of
Malacca with its prominent hill was very evident.
As the matrix clause of this complex sentence describes a specific past
situation, the addressee cannot here as in the case of (13) assume that the
understood subject form of the subclause is a pronoun with generic reference.
The intended subject form appears to be a pronoun with specific reference (/,
we or they), and as there is nothing in the matrix clause from which such a
pronoun can be inferred, the participle clause must be considered wholly
unattached and unacceptable.
In general, an unattached adverbial clause is acceptable only if (i) it serves
as a disjunct; or (ii) if the implied subject is the whole of the matrix clause;
or (iii) if the implied subject is a generic pronoun or it as a prop word, as in
the following three examples, respectively (cf. Greenbaum & Quirk 1990:
328):
(16) Putting it mildly, you have caused us some inconvenience.
(17) I'll tell you if necessary.
(18) When dining in the restaurant, a jacket and tie are required.
In closing this section, we should point out that adverbial clauses may
perform a variety of semantic functions. Some of these are illustrated in the
following examples:
(19) When independent publications were finally legalized, dozens of new titles
sprang up. (TIME)
(20) You can sit where you like. (PLACE)
(21) I left my wife because I realized I had made an awful mistake. (REASON)
(22) Shortly after the shooting, anonymous callers telephoned TV stations to warn
that more police officials would be harmed. (PURPOSE)
Conditional clauses 269
(23) Though my car is quite old, it's still in running order. (CONCESSION)
(24) I slammed down the receiver, furious after a day of frustrations.
(ATTENDANT CIRCUMSTANCE)
(25) I'll write it if somebody wants it. (CONDITION)
8.9. Conditional clauses

One particularly important subtype of adverbial clause is the one expressing
a condition. As illustrated by the following examples, a conditional clause
realizes an adverbial in a conditional sentence:
(1) If we are attacked, we will defend ourselves.
(2) You will sleep better if you get a new mattress.
A conditional sentence consists of two parts, an //-clause (the protasis) and a
matrix clause (the apodosis). By means of such a sentence the speaker
informs the hearer that the situation described by the subclause is
hypothetical and that the truth of the situation described by the matrix clause
is dependent on that of the subclause.
Typically, the two situations described by a conditional sentence are linked
both causally and temporally. In examples like (1) and (2) the situation
described by the matrix clause is not only dependent on that described by the
subclause, it also follows it in time. When in such cases the predicator of the
subclause is realized by a present verb form, that of the matrix clause is
realized by the future verb form (as in the examples) or by a modal verb form
like MAY + V (as in You may sleep better if you get a new mattress).
In conditional sentences the predicator of the subclause may also be
realized by a past verb form without past time meaning. In that case the
condition which is laid down is not neutral as it is if the verb form of the
subclause stands in the present but is biassed towards a negative response.
While it is a completely open question whether the situation described by the
subclause of e.g. If she asks him, he will be angry turns out to materialize or
not, the event described by the subclause of an example like the next one is
presented as more unlikely to take place:
(3) If she asked him, he would be angry.
In examples of this type the predicator of the matrix clause is realized by a
verb form in the past future (would + V) or by a complex verb form
beginning with a past modal form like might or could followed by an
infinitive (e.g. If she asked him, he might be angry).
Thirdly, the predicator of the subclause may be realized by a past perfect
verb form with past time meaning:
(4) If she had asked him, he would have been angry.
In such cases - where the predicator of the matrix clause is realized by a verb
form in the past future perfect (would have + V + -ed) or by a past modal
form like might or could followed by a perfect infinitive - the condition laid
down by the subclause is so strongly biassed towards a negative response
that it may be considered practically closed. The reason for this is that the
opportunity for the event described by the subclause is presented as having
already passed. Conditional clauses in the past perfect like (4) are thus nearly
always counterfactual.
Conditional clauses may begin with subordinators which are semantically
or stylistically more specialized than //, such as in case, supposing (that),
assuming (that), on condition (that) or unless (= if not):
(5) In case we are attacked, we will defend ourselves.
(6) Unless we are attacked, we will stay put.
The subordinator if may also introduce concessive adverbial clauses, which
should be kept apart from conditional clauses. This can be illustrated by an
example like If he's poor, he's at least honest (cf. Quirk et al. 1985: 1099), in
which if is synonymous with even if and even though and the situation
expressed by the subclause thus factual rather than hypothetical.
A conditional clause may be signalled not only by if οτ another sub-
ordinator but also by partial inversion (recall section 5.3.3):
(7) Should you change your mind about him, I'll invite him too.
(8) Had he been convicted of theft, he would have had to resign.
As pointed out above, the two situations described by a conditional sentence
are typically linked not only causally but also temporally, as reflected in the
fact that the verb form of the matrix clause typically refers to a stage which is
posterior in time to that referred to by the verb form in the subclause (present
- future / past - past future / past perfect - past future perfect). But there are
also conditional sentences where the two situations are not in this way linked
temporally and in which the verb form of the matrix clause is therefore not
determined by that of the subclause. This type of conditional sentence can be
illustrated by the following example where there is an implicational link
between the two events but in which the event described by the matrix clause
is not presented as subsequent to that described by the subclause:
(9) Ifquirkiness is what you crave, there's no place to beat North Korea.
In examples of this type it is often natural to insert the adverb then in the
sense of'in that case' (not 'afterwards') at the beginning of the matrix clause.
In conditional sentences where the two events are unlinked temporally and
in which it is a past form that realizes the predicator of the subclause, this
verb form has past time meaning:
Clausally realized disjuncts 271
(10) If they purchased the House of Fräser with their own money, then their funds
are the taxable funds of United Kingdom residents.
(11) But if the assault was meant to intimidate authorities, there were no signs of
anyone backing off.
Conditional sentences are not declarative exclusively but may also be inter-
rogative, imperative or exclamatory:
(12) What happens if the Queen turns out to be a foreigner!
(13) If that's all you 've got to say, let's go home right away.
(14) How wonderful it would be if that's truel
Attention should also be drawn to the existence of so-called inferential
conditional sentences, in which the speaker on the basis of the evidence put
forward in the subclause infers that the event described by the matrix clause
has taken, is taking or will be taking place. An example of this type of
conditional sentence is provided by (10) above, and it can be further
illustrated by the following sentence:
(15) If today is Friday, he is here already.
It has been pointed out (see Harder 1989, from which this example is quoted)
that the use of if - or another marker of subordination - in a conditional
sentence counts as an instruction to understand the proposition contained in
the subclause as a hypothetical basis for the speaker's message in the matrix
clause. This feature appears to be shared by all conditional sentences, i.e.
whether the two events are temporally linked or not and whether the matrix
clause is declarative or not.
8.10. Clausally realized disjuncts

As pointed out in section 4.1.2, disjuncts may be analysed as peripheral
adverbials by means of the stack convention. There are three major types of
clausally realized disjuncts: comment clauses, tag clauses and sentential rel-
ative clauses.
A) Comment clauses, which are particularly common in spoken English,
add a parenthetic comment to the content of a superordinate clause, as in
That's outrageous, I agree. They are typically markers of linguistic
interaction, as shown by the fact that their subject is in most cases realized
by a 1st or 2nd person pronoun (referring to the speaker and the hearer).
Commonly occurring examples are I'm sure, I'm afraid, I admit, I gather, I
dare say and you see, you know, mind you, you must admit. Comment clauses
like these are syntactically incomplete in that they contain a transitive verb
(e.g. KNOW) which is not followed by an object, or an adjective (e.g. SURE)
which is unaccompanied by the complementation it usually requires else-

where (see Quirk et al. 1985: 1113). Prosodically, comment clauses behave
largely like vocatives (see section 4.5). When placed finally or medially, they
are in most cases pronounced rapidly without prosodic prominence; but if the
speaker considers their content worth emphasizing, they are pronounced as a
separate tone unit. When placed initially, they are usually pronounced as a
separate tone unit:
(1) It's private, you see.
(2) I do, I'm afraid, rather act on impulse.
(3) Mind you, it isn't just embarrassment.
Many comment clauses are stereotyped fillers which are inserted into
running speech in order to establish informal contact with the hearer. When
the subject is realized by 7, their function is to inform the hearer of the
speaker's degree of certainty (e.g. / know /1 suppose) or of his emotional
attitude to the content of the matrix clause (e.g. I'm happy/sorry to say).
When the subject is realized by you, the function of comment clauses is
typically to catch or keep the hearer's attention or to request his agreement
(e.g. you see /you must admit).
There are also comment clauses whose subject is not realized by 7 or you.
The most important of these are ίο-infinitive clauses, -ing participle clauses
and finite clauses introduced by as:
(4) I don't know, to be frank.
(5) There are 200 students in the auditorium, roughly speaking.
(6) They're out, as it happens.
As, it should be added, may also introduce finite comment clauses which
have you or 7 as their subject. This can be illustrated by examples like as you
know and as I've told you. Similar examples with if are If you see what I
mean and 7/7 may say so.
B) Tag clauses. Like comment clauses, tag clauses are characteristic of
spoken English and are markers of linguistic interaction. A tag clause
consists of a subject pronoun preceded by a primary verb or modal auxiliary
(has she, did you, can't we, isn't it, etc.). It is appended to (tagged on to)
another clause, and this clause - the matrix clause - may be declarative,
imperative or exclamatory:
(7) This has been the tendency, hasn't it?
(8) Shut the door, will you?
(9) How well she sings, doesn't she!
Clausally realized disjuncts 273
The subject form of a tag clause is co-referential with that of the matrix
clause (in imperative sentences with the understood subject ioimyou), and in
declarative and exclamatory sentences the finite verb form of the matrix
clause is repeated in the tag clause or is replaced by a form of DO.
Tag clauses are also called tag questions because they signal a questioning
attitude (like their nearest equivalents in French and German n'est-ce-pas and
nicht wahr). When a tag is added to a declarative clause, the speaker simul-
taneously tells and asks the hearer the same thing. However, the questioning
force of a tag varies with its intonation. If the tag is pronounced with rising
intonation, the utterance in which it occurs is a more genuine request for
information than if it is pronounced with falling intonation. In either case the
speaker puts forward the view that the proposition contained in the matrix
clause is true, but if the tag is rising he is less certain of this view than if it is
falling:
(10) She's a 'Roman * Catholic / Jsn't she?
(11) She's a 'Roman 'Catholic / 'isn't she?
A declarative sentence ending in a tag clause is thus closer to a pure question
(e.g. Is she a Roman Catholic?) if the tag is pronounced with rising
intonation than if it is pronounced with falling intonation. In the former case
the speaker is genuinely invited to verify the truth of a statement, in the latter
only to confirm what the speaker feels reasonably sure of (see Quirk at al.
1985: 811). In fact, the main function of a tag pronounced with falling
intonation is not to request information but to make sure that the contact
between speaker and hearer is retained.
Declarative sentences ending in tag clauses are normally characterized by
reversed polarity: if the declarative sentence is positive, the tag is negative
(as in This has been the tendency, hasn't it?), and if it is negative, the tag is
positive (as in This hasn't been the tendency, has it?). However, both the
declarative clause and the tag clause may be positive. In such cases - in
which the tag is pronounced with rising intonation - the illocutionary force
of the sentence is that of a pure question (see Nässlin 1984):
(12) She's a 'Roman' Catholic / ,is she?
What we find here is illocutionary conversion, i.e. the tag clause functions as
an illocutionary converter (statement to question). The speaker is not really
telling the hearer anything but only asks whether the content of the matrix
clause is true or false.
Sentences like the following in which both the declarative clause and the
tag clause are negative are rare:
(13) She can't come, can't she?
In cases like this it would seem that the function of the tag is to challenge a
view held by the hearer. For example, it would be natural for a speaker
uttering (13) to continue his turn with an utterance like Oh yes, she can. Such
a challenging attitude is occasionally also noticeable if a declarative clause
and a tag clause are both positive. Thus an utterance like You'd like it, would
you? may in some contexts function not as a genuine question (= Would you
like it?) but as a remonstration against something intended by the hearer.
When a tag is added to an imperative clause, it is usually pronounced with
rising intonation:
(14) Shut the door, won't you?
(15) Make up your mind, would you?
By means of such imperative sentences the speaker simultaneously instructs
and asks the hearer to perform a certain action, i.e. the illocutionary value is
here a mixture of a directive and a question. Note that only a limited number
of tags can be added to an imperative clause: the subject is nearly always
realized by you and the predicator by will/would, can/could or shall/should.
When a tag clause is added to an exclamatory clause, it is nearly always
pronounced with falling intonation:
(16) How 'well she sings / "doesn't she!
Here the illocutionary value of the utterance is that of an exclamation in
which there is a relatively weak admixture of a question. That it is the
exclamatory function which dominates in such examples appears from the
fact that the tag clause cannot readily be uttered with rising intonation.
C) Sentential relative clauses. A sentential relative clause refers back to the
preceding clause, from which it is separated by intonation or - in written
English - punctuation (comma or sometimes full stop). It nearly always
begins with the relative pronoun which:
(17) The twins don't look alike, which puzzles me.
(18) The terrorists have claimed responsibility for the bomb blast, which is
exactly what we've been expecting.
In such examples, which has demonstrative-like meaning (= 'and this1).
The antecedent of a sentential relative clause need not be the entire pre-
ceding clause but may be limited to the predicate or predication of that
clause:
(19) She commutes between Boston and New York, which I wouldn't be able to
stand.
(20) She will commute between Boston and New York, which I wouldn't be able
to stand.
Polarity in complex sentences 275
The antecedent may also be a longer stretch of speech than the preceding
clause. In the following example it is constituted not only by the clause
immediately preceding which but also by the initial conditional sentence:
(21) If you aren't registered to vote, then they can't pursue you for the poll tax -
that is the received wisdom. Which left us with the possibility of one of the
ripest political ironies.
The examples examined so far show anaphoric sentential relative construc-
tions. Occasionally we come across cataphoric cases with what, as in the
following examples where the two types are juxtaposed:
(22) She was late, which was bad, but what was worse, she didn't apologize.
Which is used cataphorically after a coordinating conjunction from which it
is separated by means of a comma or some other device clearly marking the
relative clause as a parenthetical insertion, cf. also section 11.3.3 [B.c] and
Schibsbye (1970: 253), who provides the following example:
(23) Change of meaning may also be effected, by means of figurative language,
or, which is a similar process, the use of a concrete term for an abstract
conception.
Sentential relative clauses may also be introduced by when preceded by a
preposition or - in formal style - by whence or whereupon (see Quirk et al.
1985:1119f):
(24) The price of bread rose sharply, since when many have found it difficult to
make ends meet.
(25) One of the pupils stabbed another, whereupon the headmaster immediately
called the police.
In an example like Margaret Thatcher is now a life Baroness, which
everyone knows, we can replace which with as with virtually no change of
meaning. But unlike which, as is not generally used as a relative but as a
conjunction. Note also that as everyone knows is positionally less restricted
than which everyone knows: it could also be placed initially or medially. We
therefore do not classify such an os-clause as a sentential relative clause but
as a comment clause.
8.11. Polarity in complex sentences

In complex sentences negation may affect the matrix clause, the subclause or
the entire sentence. These three possibilities are illustrated in the following
examples:
(1) Sanctions don't challenge vital interests, as it happens.
(2) Police know that the shooter was not a professional.
(3) I didn't write the book in order to make money.

In (1) the comment subclause realizes a disjunct and falls outside the
semantic scope of negation (see section 7.5.8). In (2), conversely, it is only
the object subclause which falls inside the scope of negation, i.e. this
sentence can be analysed semantically as [Police know [not [the shooter was
a professional]]]. In (3), finally, both the matrix clause and the adverbial
subclause fall inside the scope of negation, i.e. negation is here biclausal:
[Not [I wrote the book in order to make money]].
In some cases - particularly in informal style - negation is transferred (or
raised) from a subclause where it belongs semantically to the matrix clause.
The matrix clause verbs which permit such negative raising are verbs of
opinion like BELIEVE, EXPECT, IMAGINE, SUPPOSE, THINK and verbs of
perception like APPEAR, SEEM, FEEL/LOOK/SOUND as i/(see Quirk et al.
1985: 1033):
(4a) I don't believe/think it's raining any longer.
(5a) I don't expect/imagine/suppose I'll pass the exam.
(6a) They didn't appear/seem to be convinced by the argument.
(7a) It doesn't look/sound as if Major knows the answer to this.
These examples should be compared with:
(4b) I believe/think it's not raining any longer.
(5b) I expect/imagine/suppose I won't pass the exam.
(6b) They appeared/seemed not to be convinced by the argument.
(7b) It looks/sounds as if Major doesn't know the answer to this.
in which the negative element is placed where it belongs semantically.
Complex sentences with transferred negation are practically synonymous
with complex sentences in which the negative element is retained in the
subclause. A sentence like It doesn't look/sound as if Major knows the
answer to this has virtually the same meaning as It looks/sounds as if Major
doesn't know the answer to this. If a difference in meaning can be detected, it
involves the force of negation, which tends to be slightly weaker in sentences
with raising (see Quirk et al. 1985: 1033).
PART III
9. Verbals
9.1. Introduction
As pointed out in section 3.3.1, we use the form term Verbal' as a cover term
for single verbs and verb groups. By a verb group we understand a group
which has a full verb as its head, for example may have been dancing, in
which dancing functions as head and the preceding words as dependents. In
verb groups with premodification like this one, all the constituents are
realized by verbs, i.e. by words that can typically be inflected for the
present/past distinction (dance/danced, have/had, etc.). As pointed out in
section 3.2.1, the verbs preceding the head of a verb group like may have
been dancing are called auxiliary verbs. There may also be dependents after
the head of a verb group, but only if the group realizes a complex predicator
such as e.g. The analysis threw up several surprises (cf. section 4.3). Verbals
usually function as predicators and may be finite or nonfmite. In finite
verbals the first verb is inflected for the present/past distinction (cf. section
3.2.1). In nonfinite verbals a dependent preceding the head may be realized
not only by a verb (as in Having given him a light, I set fire to his
moustache) but also by the infinitive marker to (as in To take a walk here
would be foolish).
Verbals typically describe situations, and in this respect they differ from
e.g. nominals and pronominals, which typically express the participants
involved in situations. In an example like Our sales representative gave your
husband the wrong tickets, the verb gave is used to describe a past time
situation where somebody hands over something to somebody and the three
noun groups are used to specify the participant roles AGENT, BENEFI-
CIARY, AFFECTED. While a sentence describes a situation and in so doing
identifies the entities performing the participant roles involved, a verbal
describes a situation without identifying these entities.
The type of situation described by a sentence is determined not only by its
lexical verb but also by the inflectional form of this verb, by syntactic
relations (e.g. the presence or absence of an object) and even by the
extralinguistic context (cf. section 7.2.3). While an objectless sentence like
He was reading describes a self-contained situation, a sentence with an
object like He was reading the report describes a directed situation, i.e. a
situation progressing toward a terminal point. In both sentences, to be sure,
the verb group describes a dynamic situation, but the subtype of dynamic
situation involved is presented in different ways.
278 Verbals
The main communicative function of a verbal - to describe a situation - is

composite in that it involves a number of lower-level communicative
functions, such as locating a situation in time, presenting it as hypothetical
rather than factual (e.g. may know) and presenting it as being in progress
(e.g. is drowning). It can therefore be thought of in terms of a functional
domain, by which we understand a general main function comprising a
number of subfunctions. In this chapter we examine the ways in which
verbals and their constituents occupy this domain, i.e. in what ways they
enable the speaker to describe situations.
9.2. Verb forms

Most lexical verbs in English have four distinct forms: a base form, an -s
form, an -ing form and an -ed form. A verb like FISH, for example, has the
forms fish, fishes, fishing, fished. A lexical verb is considered morph-
ologically regular if both the past form and the past participle form are
formed by adding the suffix -ed to the base form.
In many irregular verbs the past participle form differs from the past form
in ending in -en, as illustrated by verbs like TAKE (took - taken), BEAT (beat
- beaten) and SHOW (showed - shown/showed). For this reason - and in
order to have separate terms for the two forms - some grammarians refer to
the past participle form as the -en form.
Irregular verbs may have five, four or three inflectional forms. This can be
illustrated by respectively DRIVE (drive - drives - driving - drove - driven),
HANG (hang - hangs - hanging - hung (- hung)) and PUT (put -puts -putting
(-put-put)). The verb BE is idiosyncratic in having eight distinct forms (be -
am/are/is - being - was/were - been). This proliferation is due to the fact that
historically its forms are derived from three different roots.
There are about 200 irregular verbs in English, the exact number
depending on whether verbs with prefixes such as OUTBID, OVERSLEEP and
UNBIND are included in the list of irregular verbs or are excluded because
the list contains the corresponding verbs without prefixes (BID, SLEEP,
BIND, etc.). For an overview of the inflectional morphology of irregular
verbs the reader is referred to good dictionaries of English, such as Oxford
Advanced Learner's Dictionary.
Let us briefly consider what the different forms of a verb are used for. The
base form is used in present indicative constructions (except in the 3rd
person singular), imperative constructions, subjunctive constructions and
infinitive constructions (for the terms 'indicative' and 'subjunctive', see
section 9.8). Examples: / wash my hands in Pears soap / Wash your hands /I
insist that he wash his hands /1 would like to wash my hands. The -sform is
Verbforms 279
used in 3rd person singular present indicative constructions (He washes his
hands in Pears soap). The -ing form is used in progressive constructions and
in participle constructions (She is washing her hands /He hates washing his
hands). The -edform, finally, is used partly for the expression of pastness,
partly participially for the expression of the perfect and the passive.
Examples: / washed my hands this morning / I've washed my hands / The
crops were washed away by the floods.
The -ing suffix is pronounced /in/ in both regular and irregular verbs, as
illustrated by drowning Airaunin/ and singing /"sirjirj/. In some regional and
social varieties /in/ and /irjg/ are found as alternative pronunciations. The -s
suffix is pronounced in three different ways, depending on the nature of the
final sound segment in the base form:
(i) /iz/ (in some varieties /az/) in verbs ending in a sibilant, i.e. in one of the
consonants /s z J1 g tj* dg /. Examples: kisses, buzzes, wishes, rouges,
watches, judges.
(ii) /s/ in verbs ending in a voiceless non-sibilant, i.e. one of the consonants
/p t k f θ/ (/h/ does not occur finally). Examples: hops, bets, kicks, laughs,
baths.
(iii) /z/ otherwise, i.e. in verbs ending in a vowel or a voiced non-sibilant
consonant. Examples: sees, dies, goes/begs, sings, sells.
These rules apply not only to regular verbs but also to irregular verbs (apart
from the modal auxiliaries which do not accept an -s suffix and have only
two forms, e.g. can - could). As can be seen, a supporting vowel is inserted
in those cases where the sibilant suffix consonant is added to a base form
ending in a sibilant consonant, i.e. to a closely related or identical sound.
Otherwise the suffix consonant agrees with the preceding consonant with
respect to voicing. In either case, the pronunciation of the suffix can readily
be understood in terms of 'ease of articulation*. While it would be difficult to
pronounce the final consonant clusters in e.g. /wiJV, /betz/ and /begs/ -
sound sequences which are ruled out by the phonotactic rules of English - it
is much easier to pronounce e.g. /wi J*iz/, /bets/ and /begz/.
The -ed suffix is also pronounced in three different ways depending on the
nature of the final sound segment in the base form:
(i) /id/ (in some varieties /ad/) in verbs ending in an alveolar stop consonant,
i.e. in /t/ or /d/. Examples: heated, handed.
(ii) /t/ in verbs ending in a voiceless consonant other than /t/. Examples:
stopped, watched, kicked, laughed, bathed (in BrE in the sense of 'gave a
bath to'), kissed, wished.
280 Verbals
(iii) /d/ otherwise, i.e. in verbs ending in a vowel or a voiced consonant other
than /d/. Examples: kneed, died, glowed / begged, hanged,felled.
As in the case of the -s suffix, the three pronunciations of the -ed suffix can
readily be explained phonetically: a supporting vowel is inserted for ease of
articulation if the final sound segment is of the same specific articulation
type as the suffix consonant; otherwise the suffix consonant assimilates to
the preceding sound with respect to voicing.
Orthographically, the -5 suffix has a variant -es, which occurs if the base
form of the verb ends in a sibilant or in a single written o, cf. examples like
-wish/wishes and go/goes. Conversely, the -ed suffix has a variant without e
which occurs if the base form ends in the letter e, as illustrated by
knee/kneed^ referee/refereed and please/pleased.
Apart from this orthographic variation in the suffix, it should be noted that
the addition of a suffix may bring about a change of spelling in the base
form. If -ing is added to a verb ending in 'mute1 e, this letter is usually
dropped, cf. examples like live/living and fake/faking. As shown by e.g.
age/ageing and dye/dyeing, however, there are exceptions to this ortho-
graphic rule. If -ing is added to a base form ending in ie, secondly, this letter
sequence is replaced by y, as in die/dying, lie/lying and tie/tying. The spelling
of a verb may also be affected by -s and -ed. If either of these suffixes is
added to a base form ending in ay preceded by a consonant, this y is changed
to i(e), for example in try/tries/tried (but not in e.g. play/plays/played where
y comes after a vowel letter).
A final consonant may also be doubled before -ing and -ed. This happens if
the base form is monosyllabic and its final consonant is preceded by a vowel
spelled with one letter, for example in pat/patting/patted (but not in e.g.
sweat/sweating/sweated), and it also happens if the base form is polysyllabic
and has stress on the last syllable, for example in propel/propelling/pro-
pelled. In BrE, but not usually in AmE, consonant doubling is further found
in some polysyllabic verbs whose last syllable is unstressed and ends in / or
m. Examples illustrating this are travel/travelling/travelled and program/pro-
gramming/programmed. In polysyllabic words whose last syllable is un-
stressed and ends in p there is sometimes doubling but usually not (compare
worship/worshipping/worshipped with develop/developing/developed). In a
few verbs there is vacillating orthography, for example in focus/focus(s) ing/
focus(s)ed. Finally it should be mentioned that if a base form ends in c, there
is 'doubling' in the shape of ck (as in panic/panicking/panicked).
9.3. The external relations of verbals

Finite verbals always function as P:
The internal structure of verb groups 281
(1) He had always loved Rosemary.

but may in that capacity be coordinated, in which case they technically
function as CJTs within the P function:
(2) Roger bought and sold companies.
Nonfmite verbals also function as P and CJT within P:
(3) Having always loved Rosemary, he moved to Palmer.
(4) She let Roger buy and sell companies.
but they may assume other clause functions:
(5) To love is more important than to work. (S)
(6) To negotiate at this point would be to surrender. (C)
(7) I do not want to go. (O)
(8) He stopped to smoke. (A)
In the following examples the italicized verb serves as DEP:
(9) By leaving he indicated his dissatisfaction with the negotiations.
(10) There was a dancing girl on the stage.
Verbals realizing other functions than P still have P potential in full nonfinite
clauses realizing the same functions in the main clause, e.g.:
(51) To love her is more important than to work.
(8') He stopped to smoke a cigar.
(91) By leaving the meeting he indicated his dissatisfaction with the negotiations.
9.4. The internal structure of verb groups

In verb groups with more pre-head dependents than one, the order of
auxiliaries is fixed and looks like this:
Modal Perfect Progressive Passive
can
may
must have be be
shall
will
This ordering can be illustrated by the following very rare example with four
auxiliaries, repeated from section 7.4.4:
(1) ... another £98 m was paid direct from absent parents to their former parents.
Much ofthat could well also have been being paid before April last year, Mr
Field argued.
282 Verbals
Each auxiliary determines the inflectional form of the following verb. A

modal is followed by a base form, perfect HAVE by an -ed form, progressive
BE by an -ing form and passive BE by an -ed form: e.g. may call, has called,
is calling and is called.
In nonfinite verb groups there are no modal auxiliaries, such auxiliaries
having only finite forms. In participial verb groups we find the forms having
+ V + -ed, being + V + -ed, having been + V + -ed and having been + V +
-ing. Examples (cf. Davidsen-Nielsen 1990: 153f):
(2) Palme accused him of having made "contact with American spies".
(3) The Navy conducted 2,000 patrols without being spotted by the Soviets.
(4) Her arm was in a sling but showed no signs of having been damaged.
(5) Having been drinking since last night I feel somewhat indisposed.
The absence of a non-perfect progressive participle construction is due to the
fact that English does not allow two consecutive -ing forms (see Palmer
1987: 34). The occurrence of three auxiliaries cannot be ruled out entirely.
But though an example like Having been being pestered with phone calls all
morning, I'm in a nasty mood is not ungrammatical, it is much less likely to
occur than a corresponding sentence without being.
In infinitive verb groups we find the forms to have + V + -ed, to be + V +
-ed, to be + V + -ing, to have been + V + -ed and to have been + V + -in:
(6) Roger was thought to have resigned.
(7) John Paul was said to be angered by the editorial.
(8) He appears to be working for the Russians.
(9) The meeting seemed to have been planned by the CIA.
(10) Linda appears to have been walking in the rain.
As in participial verb groups, the occurrence of three auxiliaries cannot be
ruled out altogether. But though a constructed example like / believe him to
have been being operated on all morning is not ungrammatical, it is so
cumbersome that it is highly unlikely to occur.
Nonfinite verb groups realized by an infinitive verb group without to are
found in sentences with had better, unless this modal idiom is analysed as an
auxiliary. Examples illustrating this are She'd better have finished the report
soon and You 'd better be studying hard when I return.
9.5. Auxiliaries and their delimitation

Unlike a lexical verb, an auxiliary typically requires the presence of another
verb. Auxiliaries are also called grammatical verbs because they perform the
same kinds of function as verbal inflections, which are indisputably
Auxiliaries and their delimitation 283
grammatical entities. For example, the work done by the auxiliary WILL in a
sentence like It will matter a great deal resembles that done by the suffix -ed
in // mattered a great deal in that it is used for the expression of time.
Auxiliaries form a closed system and are separated from lexical verbs by a
combination of morphological, syntactic and semantic criteria. They are
typically morphologically defective, they share a number of syntactic
features, and semantically they differ from most lexical verbs in expressing
highly general meanings, relating to tense, aspect, modality or voice.
The class of auxiliaries is commonly assumed to comprise primary HAVE,
BE, DO and modal CAN, MAY, MUST, SHALL, WILL (see section 3.2.1). All
the modals are morphologically defective: MUST has only one form and the
rest only two (can - could, may - might, shall - should, will - would). When
combined with another word, i.e. when used as an auxiliary, DO is
morphologically defective too, for the forms doing and done are not used in
this context. On the other hand, HAVE and BE have the same four and eight
forms as when they are used as lexical verbs, as illustrated by have called,
has called, had called, having called and to be called, am called, are called,
is called, was called, were called, (is) being called, (has) been called.
Syntactically, auxiliaries differ from lexical verbs in requiring no DO-
support in negative, interrogative and emphatically affirmative sentences (cf.
sections 3.2.1, 7.5.2 and 7.5.10).
In the case of so-called 'code-constructions', i.e. in sentences with tag
questions or similar constructions where the description of a situation is
repeated, we observe the same difference:
(1) They were laughing, weren't they?
(2) They kept laughing, didn't they?
(3) They were laughing and so was their teacher.
(4) They kept laughing and so did their teacher.
Here DO serves as a pro-form (cf. section 4.2.3).
The syntactic properties of English auxiliaries just described are often
referred to as the NICE-properties because they involve negation, inversion,
code and emphatic affirmation (see Twaddell 1965 and Huddleston 1976).
While these properties constitute a relatively practicable and reliable
criterion for establishing a class of auxiliaries, it should not be overlooked
that they are shared by BE and (partly) HAVE when these are used as lexical
verbs (see section 3.2.1). Furthermore, a modal verb like OUGHT requires no
DO-support in negative and interrogative sentences (e.g. You oughtn 't to have
said that / Ought I to see a doctor?), and in tag questions it is more often
than not repeated (compare We ought to go, oughtn't we? with We ought to
go, shouldn't we? and We ought to go, hadn't we?). In other cases of code,
284 Verbals
however, OUGHT is not normally repeated; for example, You ought to go and
so ought your wife borders on unacceptability. Another reason why OUGHT
can be excluded from the class of central auxiliaries is that it is not attached
directly to the verb it modifies but obligatorily separated from it by to. In this
respect it differs both from the central auxiliaries and from verbal inflections.
In some descriptions of English, OUGHT is regarded as a semi-auxiliary, and
so are USED TO, NEED and DARE in certain types of construction in
nonassertive contexts. An 'ideal1 auxiliary has all the NICE-properties and is
accompanied by a nonfinite verb form while an 'ideal1 lexical verb has none
of the NICE-properties and is followed by non-verbal constituents only
(except in examples of the type He tried to escape and My sister enjoyed
swimming, where a verbal constituent serves as direct object). Between the
end points of this auxiliary-lexical scale a number of intermediary types of
auxiliary can be identified, the most important of which are semi-auxiliaries
and the so-called catenatives, which we turn to below.
As semi-auxiliaries DARE and NEED show three characteristics (cf.
Schibsbye 1970: 24f, 83f): a) there is no DO-support; b) the following
infinitive is a bare infinitive; c) there is no third person singular -s in the
present; compare Dad need not be told (where need is a semi-auxiliary with
an association of 'requirement' attached to the circumstances) with Dad does
not need to be told (where need is a full verb with an association of
'requirement' attached to the subject). Blends of the two uses are not unusual
for DARE: e.g. They do not dare ask for more money (where there is DO-
support but the following infinitive is bare) and He dares not try to contact
the authorities (where DARE takes the third person singular -s suffix but is
followed by the bare infinitive).
USED TO, which expresses past states or (discontinued) habits, cannot be
ruled out as an auxiliary in negative and interrogative constructions (He used
not to work late hours / Used she to smoke?), or even in tag questions (They
used not to smoke, use(d)n't they?). More commonly, however, USED TO
takes DO-support in constructions (He didn't use to work late hours /Did she
use to smoke? / They didn 't use to smoke, did they?). The 'd' is sometimes
retained in writing (... didn't used to ... /Did ... used to ...?) though this is
generally considered incorrect. USED TO always behaves like a full verb in
other CODE constructions (e.g. He used to smoke a pipe, and so did she).
What we find in English is thus in fact a scale ranging from clear
auxiliaries to clear lexical verbs. While a verb like MAY belongs to the
former category and a verb like NEGOTIATE to the latter, verbs like OUGHT
TO, NEED, DARE, USED TO, HAVE TO and BE TO occupy a borderline area
between the two. Attention must also be drawn to so-called catenatives such
as GET and KEEP, which behave like full verbs with respect to the NICE-
Tense and aspect 285
properties but which may serve auxiliary-like functions in examples like He

got arrested (cf. He was arrested) and They kept laughing (cf. They were
laughing). In this grammar we prefer to operate with a small class of
auxiliaries proper (BE, HAVE, DO, CAN, MAY, MUST, SHALL, WILL) and to
analyse verbs like OUGHT, NEED, BE GOING TO, etc. as semi-auxiliaries
and verbs like GET and KEEP as catenatives.
Even within the class of auxiliaries there are differences with respect to the
degree to which they can be considered grammatical words. The modals have
more lexical meaning than BE, HAVE and particularly DO (which is
semantically empty). Morphologically, on the other hand, the primary
auxiliaries are less defective than the modals. Within the class of modal
auxiliaries there are differences as well, for the meaning of these is more
general in some of their uses than in others (see section 9.9). For example,
the meaning of MAY is more general in The economy may get worse (where
the auxiliary is used to judge the probability of a situation) than in May I
come in? (where it expresses permission).
We should point out that it would be possible to analyse WILL used for the
expression of future time (e.g. in The meeting will take place) not as a modal
auxiliary but as a primary auxiliary. The reason we have chosen not to do so
is that WILL belongs in a paradigm with CAN, MAY, MUST, SHALL where it
blocks out the occurrence of any of these verbs. As WILL in one of its senses
is clearly a time marker, however, we return to this verb not only in section
9.9 on modality but also in section 9.6 on tense and aspect.
In closing this section we recapitulate that an auxiliary in English is a
'helping verb1 which modifies another verb to which it may be directly
attached (and usually is), that it expresses general meaning, that it is
(typically) morphologically defective and that it requires no DO-support in
sentences involving negation, inversion, code or emphatic affirmation.
9.6. Tense and aspect

9.6.1. Introduction
Tense and aspect are closely related categories in that both of them concern
the presentation of situations. Tense is defined as grammatically expressed
assignment to situations of 'location in time' and can be illustrated by
examples like Linda lives in Stockholm and Linda lived in Stockholm. In
using the inflection -s in the first of these, the speaker instructs the hearer to
identify a situation that applies at the moment the utterance is made, and in
using the inflection -ed in the second to identify a situation that applies
before this moment. Aspect is defined as grammatically expressed assign-
286 Verbals
ment of 'situational focus1 and can be illustrated by examples like It was

snowing in Stockholm and It snowed in Stockholm. In using the auxiliary was
and the inflectional ending -ing in the first of these, the speaker instructs the
hearer to select an internal focus, i.e. to adopt an in medias res perspective
and view the situation as unfolding. In using the simple verb form in the
second example, the speaker instructs the hearer to select an external focus,
i.e. to view the situation from without, as a complete unit.
In English, tense and aspect are tightly interwoven. We therefore treat
them together and operate with a fused tense-aspect system. The meanings
belonging to this system may be expressed by a verbal inflection, an
auxiliary or a combination, as in happened, will happen and has happened /
was happening, respectively.
The tense-aspect system in English involves four ordered choices:
1. present: past
2. future: nonfuture
3. perfect: nonperfect
4. progressive: nonprogressive
The first distinction is marked inflectionally (as in happens : happened and
has : had) and is deictic. The term 'deixis' (a Greek word that means
'pointing') refers to those features of a language which are relative to the
place and time of the utterance, and which can therefore only be properly
understood in relation to the speech situation. Clear examples of deixis are
locative adverbs like HERE and THERE and temporal adverbs like NOW and
THEN (at the place/time of the speaker: not at the place/time of the speaker).
The other three distinctions are relative to the first, deictic one. The future
is signalled by a form of non-volitional WILL and the nonfuture by the
absence of this auxiliary, as in will happen : happens and would happen :
happened (cf. Davidsen-Nielsen 1988). The perfect is signalled by a form of
the auxiliary HAVE followed by an -ed verb form and the nonperfect by the
absence of this combination, as illustrated by has happened : happens and
had happened : happened. The progressive is signalled by a form of the
auxiliary BE followed by an -ing participle and the nonprogressive by the
absence of this combination, as illustrated by is happening : happens and
was happening: happened.
As there are four binary distinctions, the speaker has at his disposal a total
of sixteen tense-aspect forms: the present (happens), the past (happened), the
present future (will happen), the past future (would happen), the present
perfect (has happened), the past perfect (had happened), the present future
perfect (will have happened), the past future perfect (would have happened),
the present progressive (is happening), the past progressive (was happening),
the present future progressive (will be happening), the past future progressive
(would be happening), the present perfect progressive (has been happening),
the past perfect progressive (had been happening), the present future perfect
progressive (will have been happening) and the past future perfect pro-
gressive (would have been happening).
In our account of tense-aspect meanings we shall adopt a functional-
instructional rather than a traditional referential point of view (see Harder
(1996), by whose work this section is significantly influenced). In an
example like Arsenal scored a crucial goal we prefer to say not that the past
refers to a situation that precedes the moment this utterance is spoken, but
rather that it instructs the hearer to identify a situation before this moment as
that which the descriptive content applies to ('world-before-now'). In
examples like It smells awful and Wright passes the ball to Bergkamp,
similarly, we shall say not that the present refers to a situation that is
simultaneous with the moment of speech but rather that it instructs the hearer
to identify a situation at this moment as that which the descriptive content
applies to ('world now'). In other words, we regard finite verb forms as forms
that tell the hearer where to look, i.e. where to tag the descriptive content of a
sentence on to the world.
9.6.2. Deictic forms: present and past

The present (e.g. happens) instructs the hearer to identify a situation as it is
at the moment of speech ('world now') and the past (e.g. happened) to
identify a situation as it was before this moment ('world-before-now').
Complex verb forms also instruct the hearer to identify a situation either as it
is at the moment of speech (e.g. will happen, has happened, is happening) or
as it was before the moment of speech (e.g. would happen, had happened,
was happening). But here other instructions - which we return to in sections
9.6.3 to 9.6.6 - are given as well.
The use of the simple past can be illustrated by the following examples (cf.
Davidsen-Nielsen 1990: 124):
(1) A luxury jet yesterday brought Imelda Marcos to New York.
(2) The election got off to a lacklustre start at a relatively low-key rally.
(3) Nick Faldo, the British golfer, won the Volvo Masters in Sotogrande.
In using the past, the speaker signals that he has a particular past time in
mind and assumes that the hearer is able to infer what this time is (see
Elsness 1997). This particular time may be expressed by a temporal
adverbial which functions as an anchor (the way yesterday does in example
(1)). Otherwise it is expressed or implied by the wider linguistic context (e.g.
288 Verbals
if / hated it comes after / was in L. A. last week) or implied by the situational

context (as in Did the postman bring any letters?). A sentence in the past like
John understood - unlike a sentence in the present like John understands -
thus cannot be used appropriately unless the hearer is given a clue about the
specific occasion on which the situation described took place. If unanchored
past time needs to be expressed, the present perfect has to be used instead (as
in John has resigned, see section 9.6.4).
Modal usage of the past will be dealt with in section 9.7, and the use of the
past in indirect speech in section 9.6.13.
As pointed out in section 9.6.1, the present instructs the hearer to identify a
situation at the moment of speech as that which the descriptive content
applies to. If the situation referred to by a sentence in the present is stative
(see section 7.2.1), it spans the moment of speech, and in such cases it is
customary to speak of the inclusive present, as in the following examples
expressing attitudes, intensive and extensive relations:
(4) George believes in God.
(5) She hates all the fuss.
(6) They are in high spirits.
(7) This factory belongs to Mr Hardcastle.
If a situation described by a sentence in the nonprogressive present belongs
to the subtype of dynamic actionality termed punctual, it coincides with the
moment of speech. This so-called instantaneous present is fairly restricted.
As pointed out by Bache (1985a: 273f; 1986: 92), it is found in (broadcast)
commentaries, demonstrations, special exclamatory sentences and perform-
atives, i.e. constructions supplementing a visual experience and/or referring
to highly regulated, ritualized or ordered events:
(8) Wright passes the ball to Bergkamp.
(9) I now remove the moss on top of the soil and top up the pot with compost.
(10) Here comes the bride!
(11) I promise to be back by ten.
Apart from such cases, dynamic situations taking place at the moment of
speech require description by means of the present progressive (e.g. Right at
this moment Jack is writing an e-mail message to his boss).
Inclusion and instantaneousness (where possible) are not properties of the
present itself but of the actionality of the sentence. In uttering e.g. (4) the
speaker instructs the hearer to identify a present-time situation of 'somebody
believing in something', and as this situation is stative it will be understood
to include the moment of speech. In uttering e.g. (8), the speaker instructs the
hearer to identify a present time situation of 'somebody passing a ball to
somebody else', and as this situation is dynamic, it will be understood to

coincide with the moment of speech, i.e. be instantaneous.
The present is found also in sentences expressing habitual situations,
including:
(i) universal conditions (eternal and mathematical truths, for example):
(12) Water boils at 100 degrees centigrade.
(13) Two and two makes four.
(14) The sun sets in the west.
(ii) personal habits, i.e. constructions which describe a settled manner of
human (or animal) behaviour:
(15) Sally buys her clothes at Marks & Spencer's.
(16) I go to bed at twelve o'clock.
(17) She cycles to work.
(iii) present ability:
(18) Evelyn speaks Russian.
(19) Jim runs a mile in less than 6 minutes.
It should be remembered that the situation described by a personal habitual
construction like (15) is not dynamic instances of buying - though such
instances are implied - but rather a characterization of Sally (see section
7.2.3). Similarly, the situations described by the universal statement in (12)
and the expression of present ability in (19) are not dynamic instances of
boiling and running but rather characterizations of water and Jim.
The present virtually always counts as an instruction to the hearer to look
at the world as it is now. Note that this is the case even in examples like:
(20) The meeting takes place tomorrow.
(21) Peter tells me you're going to the States.
Both these sentences instruct the hearer to identify a situation as it is now,
the same way the semantically closely related sentences in the present future
and the present perfect do (The meeting will take place tomorrow / Peter has
told me you're going to the States). We return to the additional instructions
signalled by such sentences in sections 9.6.3 and 9.6.4.
An apparent exception to the rule that the present counts as an instruction
to identify a situation at the moment of speech is the so-called dramatic
present (or historic present), i.e. a present used to make the description of a
past situation more vivid:
(22) We proceeded along the main road. Up the road we enter the courtyard of a
rund-down palazzo.
290 Verbals
In this example the present form enter is used atypically in that it seems to
apply to the world as it was before now. But even here there is at the same
time an element of'world now', for in using the dramatic present the speaker
"steps outside the frame of history, visualizing and representing what
happened in the past as if it were present before his eyes" (Jespersen 1929:
258). In selecting the dramatic present the narrator eliminates the temporal
distance between the chain of events described and his account of it.
9.6.3. Future forms

The present future can be illustrated by the following examples (quoted from
Davidsen-Nielsen 1990: 117):
(1) In a day or two - father - you will feel yourself again.
(2) Oh dear... whatever will the Bishop say?
(3) You will not get such ham with the trappists.
The WILL found in present future forms differs from volitional WILL in
expressing time exclusively. Unlike examples like (1) through (3), an
example with volitional WILL like / will gladly help you expresses not only
futurity but also willingness. Another property of present future WILL is that
a sentence in which it occurs describes the real world categorically, though
at a time that is still ahead. In this way non-volitional WILL differs from a
modal auxiliary like MAY. While both Peter may recover and Peter will
recover describe something as still unrealized, the real world is not spoken
about categorically in the former sentence as it is in the latter. In using the
present future the speaker talks about the future as if it were certain.
Futurity does not have a structural home of its own in English as it has in
e.g. French (where it is expressed inflectionally, as in // signera 'He'll sign1)
but is a squatter in the modal paradigm (see Harder 1996: 369). Nevertheless,
it is both possible and appropriate to operate with a present future form in the
shape of non-volitional WILL + V. Note in this connection that non-
volitional WILL differs from volitional WILL not only semantically but also
syntactically in occurring in passive sentences, progressive sentences and
before HAVE + V + -ed:
(4) I will be brought back in disgrace.
(5) We'// be throwing a party.
(6) By this time tomorrow, I'// no doubt have finished sorting out the first batch
of replies.
Furthermore, WILL used for the expression of pure future differs from
modal-volitional WILL in not normally occurring in conditional or temporal
subclauses. In an example like If you'll be patient for a few minutes more, I'll
have finished we thus find modal WILL in the subclause and future WILL in
the matrix clause.
We choose to disregard the marginal realization of the present future by
SHALL + V, for the purely temporal use of SHALL found in sentences with
1st person pronoun subjects is restricted to formal BrE and by now relatively
rare. As appears from examples in which it is found, such as I feel I shall
never get over it and We shall never be as we were, this SHALL can readily
be replaced by WILL (see Davidsen-Nielsen 1990: 57).
Talking of the 'present future' (rather than merely the 'future') means that
we see it as semantically complex, i.e. its meaning is 'It applies now that
something is ahead'. Formally this can be expressed by the notation [Present
[future [situation]]]; spelled out in terms of a complex instruction, it means
that the hearer is instructed to tag the sentence on to present time ('world
now') and then look ahead at a situation. Unlike the past, the present future
does not require any expressed or implied anchor. Such an anchor is not
necessary for a proper understanding of what is meant. While a past time
anchor is required to understand a sentence like She regretted the decision,
no future time anchor is necessary for the understanding of She'll regret the
decision.
Semantically, the present future is closely related to BE going to + V. In a
large number of cases the two constructions are used for the same purpose,
i.e. to instruct the hearer to think of present time and then look ahead:
(7) It'll be very difficult.
(8) It's going to be very difficult.
While there is thus considerable overlap, there are also cases where only the
present future or BE going to can be used. For example, a sentence like It's
going to rain cannot be replaced by It'll rain if the situation described is one
where dark clouds are gathering. This is due to the fact that a function of BE
going to - but not of W I L L - is to express future of present cause.
Conversely, we find WILL but not normally BE going to in the matrix clause
of a conditional sentence:
(9) Unless something goes wrong, she'// have a baby soon.
The reason why BE going to cannot very well be used here is that in
expressing future of present cause She's going to have a baby soon means
approximately 'She's several months pregnant', and this type of meaning
obviously cannot be made dependent on a condition like that laid down by
the subclause in (9).
The most important difference between the present future and BE going to
+ V concerns scope and degree of grammaticalization (see section 7.5.8 on
292 Verbals
the semantic scope of negation). Like French ALLER found in e.g. // να venir
"He'll come', BE going to has not yet become fully grammaticalized and can
therefore only be classified as a semi-auxiliary. This appears from the fact
that it only shares some of the properties of central auxiliaries discussed in
section 9.5. As for scope, BE going to has semantic influence over a smaller
part of the meaning of a sentence than WILL. This can be illustrated by (10),
which is exceptional in containing both WILL and BE going to:
(10) She'// be going to have a baby soon (unless she changes her lifestyle).
Semantically, this sentence must be analysed as [Will [be going to [she have
a baby soon]]], i.e. WILL has semantic scope over the rest of the sentence
including BE going to.
In accounting for the present future we also have to consider some cases
where it competes with the simple present. In section 9.6.2 it was pointed out
that a sentence in the present like The meeting takes place tomorrow is
semantically closely related to a sentence in the present future like The
meeting will take place tomorrow. In both examples the hearer is instructed
to think of present time and then look ahead. In the former the additional
instruction to look ahead is given lexically exclusively by means of
tomorrow and in the latter both grammatically by WILL and lexically. While
WILL instructs the hearer simply to look ahead, TOMORROW instructs him
to look ahead to a specific point in time. In sentences of this type WILL is
therefore largely redundant and could without communicative consequences
be omitted. Nevertheless the use of the simple present in sentences signalling
aheadness is not common. In simple sentences and in matrix clauses of
complex sentences it is restricted to those cases where something planned
and certain is involved and where there is a future time adverbial. Further-
more it usually requires that the actionality of the sentence is dynamic:
(11) Mr Major visits Poland next week.
(12) President Yeltsin arrives in India tomorrow for a weekend visit.
In subclauses, however, the simple present is the rule rather than the
exception, particularly in conditional and temporal clauses:
(13) You'll sleep better if you get a new mattress.
(14) When my son comes home, we'll kill the fatted calf.
Here it may perhaps be assumed that WILL has semantic scope over the
entire complex sentence.
Futurity can also be expressed by the verbal idiom BE about to and the
modal semi-auxiliary BE to, and here as well more specific meanings are
signalled than by WILL. The former is used to describe the imminent future,
as in She's about to join the navy, and the latter can be used to describe a
future event which has already been arranged, as in There is to be a new

hearing.
Modal uses of the present future will be dealt with in section 9.7.
Let us turn now to the past future. If this form is used temporally, it
instructs the hearer to think of a time in the world before now (i.e. to think of
some past time) and then look ahead; in other words, its meaning is 'It
applied then that something was ahead'. This use is found in subclauses of
sentences with matrix clause verbs of thinking, believing, feeling or knowing
(see Davidsen-Nielsen 1990: 139):
(15) I expected it would take us three hours to reach the summit, but now I realize
I was too optimistic.
Here would take instructs the hearer to think of some past time and then look
ahead.
In indirect speech the past future is typically a so-called backshifted
version of the present future. We return to indirect speech in section 9.6.13.
In matrix clauses the past future is common in conditional sentences, i.e. in
examples like If she asked him, he would be angry. Here the matrix clause
does not describe a past time situation but a hypothetical situation that is
dependent on the truth of the situation described by the subclause and which
is temporally ahead of this. The use of the past future in conditional
sentences was dealt with in section 8.9 and will not be taken up again.
9.6.4. Perfect forms

In using the present perfect the speaker instructs the hearer to think of
present time and then to look back at a situation. The former instruction is
given by selecting a present form of HAVE and the latter by means of the -ed
participle. The meaning of the perfect is thus 'It applies now that something
is anterior in time'. This can be expressed by the notation [Present [anterior
[situation]]]. Examples (cf. Davidsen-Nielsen 1990: 127):
(1) I haven't seen Marilyn like that before.
(2) It!s become part of the folklore.
(3) I've lived in Copenhagen since 1958.
(4) A Protestant paramilitary leader has been given a 19-year sentence.
As appears from e.g. (2) and (3), the present perfect implies either a present
result or a present continuation of a previous situation. Normally the former
is the case if the actionality of the sentence is dynamic and the latter if the
actionality is stative.
Unlike the past, the present perfect does not require any expressed or
implied anchor (although it may be present, as in (1) and (3)). Such an
294 Verbals
anchor is not necessary for a proper understanding of what is meant (for

example by sentences like (2) and (4)). While a past time anchor is required
to understand a sentence like She regretted the decision, no anchor is
necessary for the understanding of She's regretted the decision.
The present perfect combines with adverbials which include the moment
of speech, e.g. up to now, so far, yet and since 1958. Such adverbials do not
normally combine with the simple past; but exceptions to this rule are found
in AmE, as illustrated by Did the children come home yet? (BrE: Have the
children come home yet?). On the other hand, the present perfect does not
combine with adverbials which exclude the moment of speech, such as
yesterday, a year ago and the other day. Adverbials of this type require the
past, which in itself signals that the speaker has a particular past time in mind
(see section 9.6.2). A third group of time adverbials which are neutral with
respect to the temporal distinction just mentioned - e.g. recently, today and
this morning - combine both with the present perfect and the simple past.
When they are used in sentences in the past, the time described is divorced
from the moment of speech. This can be illustrated by 7 saw her this
morning, spoken in the late morning, afternoon or evening of the same day.
In sentences where they combine with the present perfect, on the other hand,
no such separation from the moment of speech is signalled. This can be
illustrated by I've seen her this morning, normally spoken in the morning of
the same day.
In many cases a past situation may equally naturally be described by a
sentence in the present perfect and a sentence in the past. If the speaker's
balance of focus inclines towards the present time result or continuation of
this situation, he will use the present perfect, i.e. instruct the hearer to tag the
situation on to present time and then look back on a situation. If the focus is
on the past situation itself, on the other hand, he will select the past, i.e.
instruct the hearer to identify a past time situation directly. This can be
illustrated by I have made that point in the telegram and 7 made that point in
the telegram.
In their relation to the present perfect and the past, the time adverbials
ALWAYS, EVER and NEVER constitute a special case (see Davidsen-Nielsen
1990: 128f). As these words describe all-inclusive time, one would expect
them to go with the present perfect rather than the past, for as already stated
the use of the latter form signals that the speaker has a particular past time in
mind. The expected combination is indeed found very frequently:
(5) I have always approved of the honours system.
(6) One of the best writers we 've ever had.
(7) I 've never trusted hand-squashers.
However, the three adverbials are also common in sentences in the past. In
some of these this is because the speaker does have a particular past time in
mind, a fact which may be indicated by other time adverbials or which the
hearer understands for non-linguistic reasons:
(8) In my childhood I always detested celery.
(9) Did you ever meet John Lennon?
(10) Before the war she never wrote poems.
Here the present perfect could not be used. But ALWAYS, EVER and NEVER
are sometimes also found in sentences in the past where there is no otherwise
expressed or understood specific past time involved:
(11) James was always a man of honour.
(12) Did you ever hear of incest?
(13) I never saw such a crowd.
Examples like these are exceptions to the rule that the past counts as an
instruction to identify a situation as it was before the moment of speech, and
express the same meaning as James has always been a man of honour, Have
you ever heard of incest? and I've never seen such a crowd.
One reason why it is sensible to operate with an integrated tense-aspect
category concerns the nature of the perfect. In being to do with temporal
location (a past time situation), it may be analysed as part of a pure tense
system (as it is by e.g. Huddleston 1995). But in simultaneously dealing with
the way the speaker looks at the situation expressed (see Bache 1985a: 5ff,
124ff, 1995: 268ff), it may also be analysed as part of an aspect system (as it
is by e.g. Quirk et al. 1985). It should also be pointed out that the perfect is
tightly interwoven with the action category (see Bache 1994). In a resultative
example like Alex has turned off the telly, the present time situation is Stative
(i.e. the telly is in a state of being off at the moment of speech) while the
anterior situation is dynamic (more specifically punctual, see section 7.2.1).
In a continuative example like They've owned the house for many years, on
the other hand, both the present time situation and the anterior situation are
stative.
In using the past perfect the speaker instructs the hearer to tag the sentence
on to 'world-before-now' and look back on a previous situation. The first
instruction is given by means of the past form had and the second by means
of the -ed participle. The meaning of the past perfect is thus 'It applied at a
past time that something was anterior in time'. This can be captured by the
notation [Past [anterior [situation]]] and illustrated by examples like these:
(14) When the second half began, Rush had scored two goals.
(15) Well, I had gone off to church and everything was peaceful and still.
296 Verbals
Here scoring two goals and going off to church are represented as anterior to
the past time the hearer is instructed to reckon with.
The past perfect is common in indirect speech as a TDackshifted' version of
the past or of the present perfect (cf. section 9.6.13). This can be illustrated
by examples like Linda told me she had gone to Egypt last year and Linda
said she hadn't been to Egypt since 1990, which should be compared with /
went to Egypt las t year and / haven't been to Egypt since 1990.
The past perfect competes with the simple past in clauses beginning with
after, such as After we (had) parked, I peeped through a flap in the tent and
saw it all. As after unequivocally places the situation described in the matrix
clause as subsequent to that described by the subclause, the past perfect can
here without loss of information be replaced by the simple past. A similar
vacillation is found in matrix clauses containing an adverbial introduced by
before: I (had) read the novel before I visited my sister.
Modal usage of the past perfect will be dealt with in section 9.7.
Perfect forms are also found in nonflnite verb groups:
(16) It is better to have loved and lost than never to have loved at all.
(17) Accusations of having shirked one's responsibilities are difficult to face.
Here the time the hearer is instructed to look back from is unspecified.
What is common to any perfect form is that the situation described is
anterior to a time of reckoning (see Harder 1996: 382). The time of
reckoning is at the moment of speech in examples (1) through (7), before the
moment of speech in (14) and (15) and unspecified in (16) and (17).
Note finally perfect-like constructions like the following with BE rather
than HAVE:
(18) The guests are gone.
The difference between this sentence and an ordinary perfect (The guests
have gone) is slight: (18) expresses pure stative meaning ('they are not here')
whereas the example with HAVE expresses this state as a result of the prior
situation of'going'.
9.6.5. Future perfect forms

The meaning signalled by the present future perfect can be expressed by the
notation [Present [future [anterior [situation]]]]. In using this form the
speaker encodes a complex instruction that can be paraphrased: tag the
sentence on to present time, look ahead to a future time, and then place the
situation before that. In its bidirectionality the present future perfect is
semantically quite complex, and not very frequent. It can be illustrated by the
following examples:
(1) The committee will have finished its work on April 30th.
(2) If you could be patient for a few minutes more, I77 have finished.
(3) By this time tomorrow, I'// no doubt have finished sorting out the first batch
of replies.
By means of the first of these sentences the speaker instructs the hearer that
right now there is ahead of him a time in relation to which the committee's
finishing of their work is anterior. Here the time from which the hearer is
instructed to look back is specified by means of the adverbial on April 30th.
Adverbials occurring in sentences in the present future perfect are often
realized by preposition groups beginning with before or with by (which in its
temporal use means 'not later than1 and is therefore semantically close to
before) and whose complement is realized by future time expressions like
tomorrow and next week. As before and by signal anteriority and tomorrow,
next week, etc. signal futurity, the prevalence of adverbiale of this type in
sentences in the present future perfect is hardly surprising.
In temporal subclauses, future anterior situations are expressed by the
present perfect rather than by the present future perfect (cf. also section 9.6.3
on the use of present nonfuture forms about the future in temporal
subclauses), compare:
(4) 177 have finished when you come.
(5) When you have finished I'll leave.
These examples also show that the sequence of events depends not so much
on the division of labour between the matrix clause and the subclause but on
the verb forms chosen: in both examples the event expressed by the perfect
form is presented as anterior to the event expressed by the nonperfect form.
The past future perfect has the semantic structure expressed in the formula
[Past [future [anterior [situation]]]]. In using it, the speaker thus instructs the
hearer to tag the sentence on to an identifiable past time, go ahead to a later
time and then place the situation before that. The following examples (cf.
Davidsen-Nielsen 1990: 141) illustrate the use of this form:
(6) I was convinced that she would have finished the book before the first of
April.
(7) I was hoping his fit of rage would have culminated soon.
The use of the past future perfect in conditional sentences, e.g. If she had
asked him, he would have been angry, was dealt with in section 8.9. Its use in
indirect speech and its modal use will be dealt with in sections 9.6.13 and 9.7
respectively.
298 Verbals
9.6.6. Progressive forms: introduction

As indicated by its name a progressive form serves the purpose of presenting
a situation as being in progress, or unfolding, i.e. with an internal focus. In
using a nonprogressive form, on the other hand, the speaker presents a
situation as a fact, or a complete unit, i.e. with an external focus. This
distinction, which is often described by grammarians in terms of the aspect
category, can be illustrated by the following examples:
(1) It was raining in Dublin.
(2) It rained in Dublin.
While (1) describes only the middle phase of the situation involved, i.e.
adopts an in medias res perspective excluding both the initial phase and the
terminal phase, (2) describes the situation without concern for its internal
phasal constituency. The difference between presenting a situation with an
external focus (i.e. with a nonprogressive form) and presenting it with an
internal focus (i.e. with a progressive form) represents the fourth choice in
our integrated tense-aspect system in English (after present/past, future/non-
ruture and perfect/nonperfect).
In accounting for progressive forms it is essential also to take actionality
into consideration (see section 7.2). At the most general level it should be
pointed out that the progressive is normally used only in sentences whose
actionality is dynamic. That it is uncommon in sentences describing Stative
situations is not surprising, for progression entails an input of energy that
characterizes dynamic but not stative actionality. For example, sentences like
Ottawa is the capital of Canada (intensive), The village lies in a dark valley
(extensive) and George believes in God (attitude) could not be changed to
corresponding sentences in the progressive. Note secondly that though e.g.
My back aches and The wound itches can be changed to My back is aching
and The wound is itching, the situation described is no longer viewed as
purely stative (perception) but as dynamic (self-contained). Attention should
also be drawn to examples like Sally was being silly and You're being
pigheaded where the main verb involved is BE. These differ from Sally was
silly and You're pigheaded (both intensive) in describing a dynamic situation,
i.e. being here means approximately 'acting1 or 'behaving'. In such cases
temporariness is expressed in addition to progression, and so it is in e.g.
Peter is/was living in London, which differs from Peter lives/lived in London
(extensive) in expressing non-permanent residence.
Certain verbs have a strong potential for expressing stative meaning, such
as BELIEVE, BELONG, CONTAIN, KNOW, MEAN, POSSESS and OWN, and
these do not normally take kindly to the progressive form:
(3) He possessed/*was possessing a certain wildness.

(4) This bottle contains/* is containing two pints of milk.
(5) In those days I knew/* was knowing him well, of course.
When such verbs are used dynamically to denote progression or a change of
state, they are of course compatible with the progressive form:
(6) He said it with that smug look that had been possessing him lately.
(7) The jug inexplicably seemed to be containing less water as the experiment
progressed. (Lauridsen 1986: 21)
In using a progressive form the speaker instructs the hearer to tag the
sentence on to a time (present, past, present + anterior, past + anterior, etc.)
and then to look at a progressing situation which is simultaneous with it. The
latter instruction is given by the -ing participle and the former by the
remainder of the verb group (is, was, has been, would be, etc.). Examples:
(8) The small figure is moving slowly along the fence.
(9) I'm speaking not just from the pulpit but from experience.
(10) The waitresses were looking at us.
(11) You're turning your back on people in need.
(12) The Yucatec Mayan language is disappearing among the Mayan population.
(13) I knew I was making a mistake.
In these examples the hearer is instructed to tag the sentence on to present
time (8, 9, 11, 12) or a past time (10, 13) and then to view the situation as
progressing simultaneously with it. And in an example like You'll be turning
your back on people in need, the hearer is instructed to tag the sentence on to
present time, look ahead to a future time and view the situation as
progressing simultaneously with that (see section 9.6.3).
If the subtype of actionality is self-contained as in (8), (9) and (10), the
situation progresses in a uniform way (see section 7.2.2). If it is directed as
in (11), (12) and (13), on the other hand, the situation progresses towards a
different state of affairs. This explains why sentences with directed
actionality in the progressive may be used to describe a future (posterior)
situation - usually under the control of the person referred to by the subject
form - which is in preparation at the time identified by the form of BE:
(14a) Linda is moving to France tomorrow.
(14b) Linda was moving to France the following day.
In these examples the adverbial informs the hearer that the speaker has a
situation ahead in mind; but even in the absence of such adverbials a
sentence in the progressive may be used intentionally to describe a posterior
situation which is in preparation at the time identified. Depending on the
300 Verbals
larger linguistic and/or situational context examples like Linda is moving to

France and Linda was moving to France may thus describe a situation that is
in progress at the time identified or which is ahead of it.
The subtype of dynamic situation described by a sentence in the pro-
gressive may not only be self-contained as in (8), (9) and (10) or directed as
in (11), (12) and (13) but also iterative:
(15) The telephone is/was ringing.
On the other hand the actionality of a sentence in the progressive cannot be
telic for, as mentioned in section 7.2.2, telicness requires that a natural
terminal point is included beyond which no further progression is possible.
Such a terminal point is clearly not included in e.g. Jack is/was fixing the old
motorbike. Nor can the actionality of a sentence in the progressive be
punctual, for lack of extension in time is incompatible with progression. In
e.g. She is/was switching from Spanish to English the situation described is
therefore not punctual but directed.
There are cases where a situation may equally well be described by a
progressive sentence and a nonprogressive sentence. This can be illustrated
by the following examples (see Bache 1986):
(16a) We celebrated Stephanie's birthday at my uncle's place.
(16b) We were celebrating Stephanie's birthday at my uncle's place.
Here it is the same (self-contained) situation which is described but in
slightly different ways (external vs. internal focus). Note in this connection
verbs like STAND, SIT, LIE, HOLD, KEEP, OCCUPY, SLEEP, STAY, WAIT,
WEAR. These verbs, which have a clear Stative potential, are often used to
refer to temporary posture or conditions: e.g. He waited for her in the library
/She wore her mother's wedding dress. In connection with subjects referring
to volitional agents more or less 'in command' of what is going on, such
situations are in a grey zone between dynamic and stative and are therefore
best classified as self-contained (i.e. the most stative of the dynamic
subsituations). They often permit expression by both the progressive and
nonprogressive verb forms with only a slight difference of aspectual meaning
(cf. He was waiting for her in the library / She was wearing her mother's
wedding dress).
In examples like the next ones, on the other hand, there is a marked
semantic difference:
(17a) Walter moved to the door.
(17b) Walter was moving to the door.
Here again a situation is described with an external or with an internal focus
(this is indeed the constant difference between the nonprogressive and the
progressive). But in choosing the latter focus (example (17b)) the speaker
eliminates the completion of the situation from the reference of the verbal,
i.e. he describes a situation where the door is approached but not reached. As
a result the situation is described not as telic - as it is in (17a), which
indicates that the door was reached - but as directed. The difference between
a progressive and a corresponding nonprogressive sentence is thus much
stronger if there is a concomitant difference in actionality than if the
actionality remains the same. Note in this connection that while (16a) is
entailed by (16b) - i.e. the truth of (16a) follows from that of (16b) - (17a) is
not entailed by (17b): if we were celebrating Stephanie's birthday it follows
that we also celebrated it; but if Walter was moving to the door it does not
follow that he also moved (all the way) to the door.
The choice between progressive and nonprogressive forms is basically
non-deictic but like the choices involving the future and perfect forms it
sometimes has deictic implications. In choosing the progressive (along with
the present, nonfuture and nonperfect) in a sentence like He's speaking like a
professional, the speaker relates the situation described more precisely to the
present moment than in choosing the nonprogressive in He speaks like a
professional·, while the latter sentence describes a situation which applies
generally, a habit, the former describes a dynamic situation taking place here
and now, i.e. specifically at the time and place of the speaker.
As indicated at the beginning of this section, progressive forms are often
assumed to differ from corresponding nonprogressive forms in terms of
aspect, and one reason for that is undoubtedly that the choice between
progressive and nonprogressive is basically non-deictic. However, like the
perfect forms and the future forms, progressive forms are clearly governed
by, and integrated with, the primary deictic choice between present and past.
The four ordered choices in the English tense-aspect system (present/past,
future/non-future, perfect/nonperfect and progressive/nonprogressive)
represent a cline of temporal meaning. Only the first choice (present/past)
involves pure deictic temporal meaning. The other choices express relative
time orientation more independently of the deictic base provided by the
moment of speech (see section 9.6.1), but as we have seen, they occasionally
have deictic implications. The further away a choice is from the first choice,
the weaker and the more sporadic these implications become, and the more
other factors become important, such as actionality and aspect, which affect
the perfect/nonperfect and especially the progressive/nonprogressive opposi-
tions.
We conclude this section by offering an overview of some of the character-
istic aspectual and actional meanings associated with pairs of nonprogressive
and progressive forms:
302 Verbals
Nonprogressive <-> Progressive

Aspect
external focus <-> internal focus
Action Examples
stative <-> dynamic He speaks like a professional

He is speaking like a professional
punctual <—> iterative A door slammed behind him

A door was slamming behind him
punctual <-> directed She caught up with the others

She was catching up with the others
telic <-> directed She built a new garden shed

She was building a new garden shed
self-contained <-> self-contained They walked along the beach

They were walking along the beach
9.6.7. Present and past progressive forms

In using the present progressive the speaker instructs the hearer to think of
present time and then look at a simultaneously progressing situation. The
meaning of the present progressive is thus 'It applies now that a situation is
simultaneous and in progress'. This can be expressed by the notation [Present
[progressing [situation]]] and illustrated by an example like Federal
authorities are investigating allegations of currency violations.
As pointed out in section 9.6.2, the simple present is often used to express
habitual meaning (including universal conditions, personal habits and
ability). In such cases the hearer is instructed to think of the world now but
clearly not to view a dynamic situation as taking place simultaneously with
the moment of speech. For this purpose the speaker must use the present
progressive, and if he chooses to do so the habitual meaning disappears:
(la) The sun sets in the west.
(1 b) The sun is setting in the west.
(2a) Evelyn speaks Russian.
(2b) Evelyn is speaking Russian.
(3a) Sally buys clothes at Marks & Spencer's.

(3b) Sally is buying clothes at Marks & Spencer's.
If the simple present is replaced by the present progressive in a sentence
describing a personal habit, as in (3a-b), there are two possibilities: either the
habitual meaning disappears and a situation is presented as being in progress
at the moment of speech, or the habitual meaning is retained but with the
difference that it is a temporary habit which is now described. The special
use of the present progressive for a temporary habit can be further
exemplified by sentences like I'm walking to work this week and She's
sleeping away her days. It should be added that the temporary implication of
the habitual progressive may be cancelled by adverbials expressing all-
inclusive time like ALWAYS and FOREVER, as illustrated by She's
always/forever asking silly questions. In such cases annoyance is often
expressed. Temporariness is a natural implication of progressive meaning
and thus present also in most nonhabitual progressives.
In using the past progressive, the speaker instructs the hearer to tag the
sentence on to a past time and then to look at a simultaneously progressing
situation. The meaning of the past progressive is thus 'It applied at a past
time that a situation was simultaneous and in progress'. This can be
expressed by the notation [Past [progressing [situation]]] and illustrated by
examples like the following:
(4) I thought he was simply babbling, but suddenly he got up and left.
(5) Since it was snowing, I went with her in the direction of the Praga Bridge.
(6) Like the damned in hell, I was being tossed from fire to ice. (passive)
In complex sentences a past progressive form in one clause may have the
effect of surrounding a situation described by a simple past form in another
clause by a temporal frame:
(7) When Mrs Moore returned, her husband was painting the view from their
hotel window.
(8) While I was eating, a new customer entered the restaurant.
Here the situation described by the matrix clause in (7) and the subclause in
(8) are presented as being in progress at the time when the situations
described by the subclause in (7) and the matrix clause in (8) took place.
Note that if the nonprogressive had been used in the matrix clause of (7), the
painting of the view would be subsequent to the return of Mrs Moore.
Though the framing effect of the past progressive is very common in
complex sentences with temporal clauses, it is by no means a constant
property in constructions with the subordinator while:
(9) While I was eating, I looked at the other customers.
304 Verbals
(1 Oa) While Minna was being measured for her clothes, I looked over her books.
Here the situations described by the matrix clauses are not framed by those
described by the subclauses but are simultaneous with them and span the
same stretch of time. In such cases it is also possible to use the past pro-
gressive in the matrix clause:
(1 Ob) While Minna was being measured for her clothes, I was looking over her
books.
Semantically, the difference between (lOa) and (lOb) is slight and only
involves a choice between a neutral external focus and a more marked
internal focus. Note, however, that if the replacement of a nonprogressive by
a progressive form affects the actionality of a sentence - which is not the
case in (lOa) and (lOb), both of which describe self-contained situations -
the difference between progressive and nonprogressive becomes quite clear:
(11 a) While Minna was being measured for her clothes, I finished my breakfast.
(finished = punctual)
(lib) While Minna was being measured for her clothes, I was finishing my
breakfast, (was finishing = directed)
In closing this section we should point out that in narration the framing
effect of the progressive typically serves the purpose of describing the
background frame against which a number of consecutive events described
by clauses in the simple past (or present) are recounted:
(12) At twelve o'clock sharp I left my flat. The sun was shining and people were
sunning themselves on benches in the park. I unlocked the car, fastened the
seat belt, started the engine and drove to the first intersection. On the pave-
ments pedestrians were strolling along. Suddenly an idea occurred to me.
Here each nonprogressive form makes the action advance in narrative time
by introducing a new time focus. The progressive forms do not perform this
function but are used to describe a situation that is simultaneous with that
described by the preceding nonprogressive form.
9.6.8. Future progressive forms

In using the present future progressive the speaker instructs the hearer to tag
the sentence on to present time, then look ahead to a future time and finally
to look at a simultaneously progressing situation. This can be captured by the
notation [Present [future [progressing [situation]]]] and illustrated by:
(1) They'// be leaving in half an hour.
(2) You're right. But in the meantime we'// be travelling by boat, and you can
always jump off a boat.
In some situations the present future progressive may be preferred to the

present future because WILL + V may out of context be ambiguous:
(3) I'// keep watch for you.
(4) ΙΊΙ be keeping watch for you.
Here WILL in (3) may be either volitional or purely temporal, and only by
means of the context is the hearer able to determine whether the
communicative function of the sentence is a promise or a descriptive state-
ment. In (4), on the other hand, WILL can only be understood to be purely
temporal. The speaker pays a price for this type of'disambiguation': she must
instruct the hearer to look not only at a future situation but at a future
situation as being in progress.
In using the past future progressive the speaker instructs the hearer to tag
the sentence on to a past time, then look ahead to a posterior time and finally
to look at a simultaneously progressing situation. This can be expressed by
the notation [Past [future [progressing [situation]]]] and illustrated by:
(5) I knew she would be asking payment for my meals.
In indirect speech the past future progressive is typically a backshifted
version of the present future progressive (cf. section 9.6.13).
The past future progressive occurs in the matrix clause of conditional
sentences, i.e. in examples of the type I'd be wandering around alone if it
weren't for Sonya. Here the matrix clause does not describe a past time
situation but a hypothetical situation (cf. section 8.9).
9.6.9. Perfect progressive forms

In using the present perfect progressive, the speaker instructs the hearer to
tag the sentence on to present time, then look back at an anterior time and
finally to look at a situation progressing simultaneously with the anterior-
present period (i.e. towards present time). This can be captured by the
notation [Present [anterior [progressing [situation]]]] and illustrated by:
(1) Has Father been talking to you?
(2) Minna, they tell me you've been looking for me.
(3) She'll suspect I've been telling you things.
Like the present perfect, the present perfect progressive does not combine
with adverbials which exclude the moment of speech, such as yesterday, a
year ago and the other day (see section 9.6.4).
In using the past perfect progressive the speaker instructs the hearer to tag
the sentence on to a past time, then look back at an anterior time and finally
to look at a situation progressing simultaneously with the anterior-past period
306 Verbals
(i.e. towards the past time). This can be expressed by the notation [Past
[anterior [progressing [situation]]]] and illustrated by examples like:
(4) Then the thing I had been dreading happened.
(5) One of the writers had been hanging around Edusha.
(6) I could tell that she 'd been talking for a long time.
Modal usage of the past perfect (progressive as well as nonprogressive) will
be dealt with in section 9.7 and its use in indirect speech in section 9.6.13.
9.6.10. Future perfect progressive forms

These forms are semantically highly complex, so not surprisingly they are
rarely encountered. The present future perfect progressive instructs the
hearer to tag the sentence on to present time, then look ahead to a future
time, then look back at an anterior time and finally look at a situation
progressing simultaneously with the future-anterior situation (i.e. towards the
future time). It can be formally represented as [Present [future [anterior
[progressing [situation]]]]] and illustrated by:
(1) The committee will have been negotiating the treaty for two months soon.
In using the past future perfect progressive the speaker instructs the hearer to
tag the sentence on to a past time, then look ahead to a posterior time, then
look back at an anterior time and finally look at a situation progressing
simultaneously with the posterior-anterior period (i.e. towards the posterior
time). This can be captured by the notation [Past [future [anterior
[progressing [situation]]]]] and illustrated by:
(2) I was convinced that she would have been pondering this problem all
morning before doing anything about it.
9.6.11. Nonfinite progressive forms

In complex sentences, progressive forms are also found in nonfmite verb
groups. Consider first examples in the nonfmite perfect progressive like the
following:
(1) Having been drinking whisky since Friday, I feel somewhat indisposed.
(2) He seemed to have been drinking.
Here the verb groups in italics instruct the hearer to look back at an anterior
time and then to look at a simultaneously progressing situation. As the time
the hearer is instructed to look back from is neither present nor past but
unspecified (see section 9.6.4), the appropriate formula in examples of this
type is [Anterior [progressing [situation]]]. Note, however, that an instruction
to tag the sentence on to present or past time which applies to the complex
sentence as a whole is given by means of the finite verb forms (feel vs.
seemed). If the speaker wishes to instruct the hearer to tag the sentence on to
present or past time by means of the verb groups in the subclause, he must do
so by choosing sentences with finite subclauses like As I have been drinking
-whisky since Friday, I feel somewhat indisposed (synonymous with (1)) and
// seemed that he had been drinking (synonymous with (2)).
Consider secondly examples in the nonfinite nonperfect progressive like
the following:
(3) He seems to be handling the affair very well.
(4) He seemed to be handling the affair very well.
Here the nonfinite verb forms instruct the hearer to look at a progressing
situation without specifying what time the situation is simultaneous with.
Again it is only by means of the finite verb that the hearer is able to figure
out whether he should think of present time (seem) or of a past time
(seemed). In this case, therefore, simultaneity is only involved indirectly. If
the speaker wishes to instruct the hearer to think of present time or a past
time by means of the verb group in the subclause, he must do so by selecting
sentences with finite subclauses like It seems as if he's handling the affair
very well (synonymous with (3)) and It seemed as if he was handling the
affair very well (synonymous with (4)).
Attention should also be drawn to examples where simultaneity remains an
unrealized potential meaning even in the context of the matrix verb:
(5) To be drowning is/was said to be a hallucinatory experience.
Here any time may in principle serve as a basis for the orientation of the
progressive as simultaneous.
Note finally that after certain verbs of perception (such as SEE and HEAR)
there is an opposition between bare infinitive and present participle
constructions which is semantically very similar to the distinction between
progressive and nonprogressive forms:
(6) I heard a small child cry/crying next door.
(7) She saw them laugh/laughing together.
Here the difference is primarily one of external focus (the infinitive) vs.
internal focus (the present participle).
9.6.12. Recapitulation
We can now recapitulate the standard notations and semantic instructions for
the sixteen forms expressing relative time in the English tense-aspect system.
308 Verbals
Instructions generally:
Present: tag on to world-now (general present time)
Past: tag on to world-before-now (past time)
Future: look ahead (future time, posteriority)
Perfect: look back (anteriority)
Progressive: look here (simultaneous progression)
Instructions specifically (as applied to the verb HAPPEN):
1. The present happens
[Present [situation]]
Tag a situation of "happening1 on to world-now.
2. The past happened
[Past [situation]]
Tag a situation of'happening1 on to world-before-now.
3. The present future will happen
[Present [future [situation]]]
Tag on to world-now and then look ahead to a situation of'happening'.
4. The past future would happen
[Past [future [situation]]]
Tag on to world-before-now and then look ahead to a situation of
"happening1.
5. The present perfect has happened
[Present [anterior [situation]]]
Tag on to world-now and then look back at a situation of'happening'.
6. The past perfect had happened
[Past [anterior [situation]]]
Tag on to world-before-now and then look back at a situation of'happening1.
7. The present future perfect will have happened
[Present [future [anterior [situation]]]]
Tag on to world-now, then look ahead to a future time and finally look back
at a situation of 'happening'.
8. The past future perfect would have happened
[Past [future [anterior [situation]]]]
Tag on to world-before-now, then look ahead to a posterior time and finally
look back at a situation of'happening'.
9. The present progressive is happening
[Present [progressing [situation]]]
Tag on to world-now and then look here at a simultaneously progressing
situation of 'happening'.
10. The past progressive was happening

[Past [progressing [situation]]]
Tag on to world-before-now and then look here at a simultaneously
progressing situation of "happening1.
11. The present future progressive will be happening
[Present [future [progressing [situation]]]]
Tag on to world-now, then look ahead to a future time and finally look here
at a simultaneously progressing situation of "happening1.
12. The past future progressive would be happening
[Past [future [progressing [situation]]]]
Tag on to world-before-now, then look ahead to a posterior time and finally
look here at a simultaneously progressing situation of'happening'.
13. The present perfect progressive has been happening
[Present [anterior [progressing [situation]]]]
Tag on to world-now, then look back at an anterior time and finally look at a
situation of 'happening' progressing simultaneously with the anterior-present
period (i.e. towards present time).
14. The past perfect progressive had been happening
[Past [anterior [progressing [situation]]]]
Tag on to world-before-now, then look back at an anterior time and finally
look at a situation of 'happening' progressing simultaneously with the
anterior-past period (i.e. towards the past time).
15. The present future perfect progressive will have been happening
[Present [future [anterior [progressing [situation]]]]]
Tag on to world-now, then look ahead to a future time, then look back at an
anterior time and finally look at a situation of 'happening' progressing
simultaneously with the future-anterior period (i.e. towards the future time).
16. The past future perfect progressive would have been happening
[Past [future [anterior [progressing [situation]]]]]
Tag on to world-before-now, then look ahead to a posterior time, then look
back at an anterior time and finally look at a situation of 'happening1
progressing simultaneously with the posterior-anterior period (i.e. towards
the posterior time).
It is important to remember that this list contains instructions reflecting the
basic semantics of the tense-aspect system. As we have seen, specific con-
structions may express derived meanings and/or have special uses,
depending on actional and aspectual properties. The English tense-aspect
system can also be represented by means of a tree diagram:
This shows that the first choice is between the present and the past, the
second choice between future and nonfuture, the third choice between perfect
and nonperfect and the fourth choice between the progressive and the
nonprogressive (the choice of nonfuture, nonperfect and nonprogressive is
symbolized by 0).
There is a deictic dine in the diagram: the lower you go, the less deixis
alone motivates the choice of verb form. At the same time other factors
become more relevant in the description, such as aspectual meaning, which
affects the perfect and especially the progressive forms (i.e. the third and
fourth choices, respectively) (see also section 9.6.4).
Note that [progressing] can only be the fourth choice. If it was assumed to
be the third, the non-progressive perfect forms could not be produced (has
happened, had happened, will have happened, would have happened). If it
were assumed to be the second choice, none of the eight perfect forms could
be produced.
Note finally that the proposed ordering of the four choices reflects the
ordering of auxiliaries in verb groups (see section 9.4). The first auxiliary is
present or past (will/would, has/had, is/was), and future WILL precedes
perfect HAVE, which in turn precedes progressive BE.
9.6.13. Tense-aspect in indirect speech

By indirect speech we understand the reporting of what an original speaker
said. This is normally accompanied by changes of tense-aspect, person and
other deictic elements (such as place references and demonstratives), as in:
(1) Peter said that his commanding officer would regard that as cowardice.
If this sentence is compared with its direct speech counterpart Peter said,
"My commanding officer will regard this as cowardice", it can be seen that
the past future form would regard is a backshifted version of will regard.
Here the past future does not necessarily instruct the hearer to look ahead
from some past time, i.e. it does not necessarily have its normal temporal
meaning (cf. Peter said that his commanding officer would regard that as
cowardice, and he may be right).
Backshifting in indirect speech is not obligatory, as demonstrated by the
acceptability of Peter said that his commanding officer will regard this as
cowardice. But it is more neutral than lack of backshifting, so if a verb of
saying is in the past, the verbs in the following subclauses are normally past
forms too (past, past perfect, past future or past future perfect), as in She said
she agreed with those who had suggested to reschedule the meeting. This so-
called consecutio temporum principle operates in conditional sentences too,
312 Verbals
though in such sentences it is rarely refrained from and not restricted to

concord of past forms.
In indirect speech a past form in a subclause following a matrix clause with
a verb of saying in the past may or may not be a backshifted form. This can
be illustrated by the following examples (which like several others in this
section are cited from Davidsen-Nielsen 1990: 147ff):
(2) Your wife said you didn't smoke but I see you do.
(3) Your wife said you didn't smoke when you were a teenager.
In (2), unlike (3), didn't smoke is a backshifted version of doesn't smoke.
Optional backshifting in indirect speech in English affects the first tense-
aspect choice (present vs. past). If the original construction contained a
present form, the possible backshifts can be summed up in this way:
Present (future) (perfect) (progressive)
I
Past (future) (perfect) (progressive)
This system yields the following specific shifts:

Present —> Past
Present perfect —» Past perfect
Present future perfect —» Past future perfect
Present future progressive —> Past future progressive
etc.
These shifts can be illustrated by the following examples:
(4) The counsellor said that blood was thicker than water.
(5) Mary said she hadn't been to the States since 1972.
(6) He said he would have finished sorting out the first batch of replies next
Friday.
(7) The contractor told me they would be working again next week.
If the original construction contained a past form, the possible backshifts can
be summed up in this way:
Past (progressive)
Past perfect (progressive)

This system yields the following specific shifts:

Past -» Past perfect
Past progressive —> Past perfect progressive
These two shifts can be illustrated by the following examples:
(8) Mary told me she had gone to California last year.
(9) Jim claimed that he had been staying at the Park Lodge over Christmas.
In indirect speech a past perfect form may thus represent either a backshifted
past form or a backshifted present perfect, and the hearer has to work out
which of the two types of backshifting is involved. In (5) and (8) the
temporal adverbials (since 1972 vs. last year) provide an answer to this
question. In (9), on the other hand, there is no such clue, so here the hearer
must decide whether it is a past progressive or a present perfect progressive
which has been backshifted by taking the wider linguistic context or the
situational context into consideration: the original statement could be either 7
was staying at the Park Lodge over Christmas' or '/ have been staying at the
Park Lodge over Christmas'.
As there is no verb form available for the backshifting of a past perfect
form, a sentence like the following is - out of context - ambiguous too:
(10) Peter told me that when the second half began Rush had scored two goals.
Here had scored may be either a backshifted version of scored or an
unshifted past perfect, i.e. the direct speech counterpart may be either When
the second half began Rush scored two goals or When the second half began
Rush had scored two goals. In the former case the hearer is instructed to
identify a past time situation and in the latter to think of a past time and then
to look back.
In closing this section we should draw attention to free indirect speech.
Free indirect speech is restricted to written language and furthermore differs
from indirect speech in that reporting clauses are omitted or parenthesized, in
that question and exclamation structures can be incorporated and in that
unshifted deictic words like here, this and now can be included. It can be
illustrated by examples like these:
(11) Presently, he reflected, it would become necessary for him to make some
excuse and steal quietly out of the other's life.
(12) Would she be able to recognize this interpretation of herself, he wondered?
Note finally cases of implicit indirect speech like the following where the
speaker enquires about the listener's conversation with a third party:
(13) 'So what time did the train leave tomorrow morning?1
314 Verbals
9.6.14. Tense-aspect in literary narrative

In fiction it is typically only the past verb forms (past, past perfect, past
future, past future perfect, progressive as well as non-progressive) which are
used, even in stories about imaginary future events. In using the past, the
writer instructs the reader to identify not a past time situation but an
imaginary situation at the stage reached in the narration. What is common to
both these types of situation, however, is remoteness from the (real) world as
it is now. The difference between them can be perceived by comparing the
following examples (cf. Bache 1986: 94)
(1) Ί climbed the steep dune yesterday, now it's your turn.'
(2) He climbed the steep dune before him hurriedly, not taking the time to
remove his shoes and socks. His panting under the effort of running uphill
seemed delicious to him; it was the taste of his renewed youth.
In the spoken utterance in (1), the situation of'climbing* is mentally distant
for the hearer, who is instructed to identify a past situation. In the written
piece of fiction in (2), the writer creates, or introduces, an imaginary
situation of climbing at the stage reached in the narration, simply by writing
it. The 'climbing' thus becomes mentally present for the reader, who is
instructed to 'witness', or envisage, the situation at this particular point in the
narration. More technically, the difference is that the past is deictic in (1)
while it is non-deictic in (2): the interpretation of the form in (2) is not
dependent on the reader's awareness of the moment of communication (the
writing of the novel). Thus in (2) but not in (1), we can replace the past
forms by present forms with little or no difference of meaning:
(2') He climbs the steep dune before him hurriedly, not taking the time to remove
his shoes and socks. His panting under the effort of running uphill seems
delicious to him; it is the taste of his renewed youth.
The difference between the present and the past in literary narrative is this:
the past establishes a fictional distance from reality (which, however, is more
or less neutralized once the reader has entered the fictional universe, i.e. has
started reading) whereas the present creates the illusion that there is no
fictional distance from reality and thus has a more dramatic potential (cf. the
dramatic present mentioned in section 9.6.2), representing a calculated
stylistic choice on the part of the writer. In both present-form and past-form
narration, however, the writer creates the situations of his fictional universe
by writing them and the reader is mentally present at the Occurrence' of the
situations narrated when reading them.
Outside literary narrative, a change from past form to present, or vice
versa, typically results in ungrammaticality (e.g. */ climb the steep dune
yesterday, now it's your turn) or in a change of temporal meaning (as in e.g. /
know/knew what I am/was doing).
There is in narration often an imitation of deixis in dialogues:
(3) Silas took another sip. Ί sent Fiona off to Berlin last week', he said.
There are also deictic-like uses of verb forms in narration even outside
dialogues - not relative to the writing of the novel, but relative to the stage
reached in the narration. Thus the past perfect and the past future forms
instruct the reader to look back and ahead, respectively, from the stage
reached in the narration (cf. Davidsen-Nielsen 1990: 70f):
(4) He was sitting beside his radio set which he Adjust switched off. It was late
at night. He had listened \o a symphony concert... Now all was silent.
(5) Ready to weep he prepared himself for bed. He would not sleep.
In using the past future perfect the writer instructs the reader to look ahead in
the narration and then to look back:
(6) Five days later Mrs Blair would have left her husband.
Similarly, when the present verb forms are used throughout a narrative (or
long stretches of it), the reader may be instructed by the use of other tense-
aspect forms to look back or ahead from the stage reached in the narration:
(7) A window smashes in one of the small bedrooms; the cause is Henry, who
has put his left arm through and down, and slashed it savagely on the glass.
(8) What is happening upstairs is something Howard will hear about later.
Note that the difference between the instructions for verb forms in literary
narrative and those for verb forms elsewhere involves only the first choice of
form (present vs. past), which is redefined semantically when applied to verb
forms in narration (most importantly in terms of loss of deictic meaning).
The choices involving 'posteriority', 'anteriority' and 'simultaneous
progression' apply equally well to fiction and non-fiction.
Literary narrative is similar to historical and biographical narrative, but in
these types of narrative it is a real past time situation the reader is instructed
to identify. Furthermore, the writer may step outside the narration and in so
doing select a present form (e.g. The next day I was appointed chairman. I
must now get down to work). The same goes for ordinary everyday narrative.
With respect to tense-aspect, historical, biographical and everyday narrative
resemble non-narrative language in that the addressee is instructed to identify
a situation in real time. But they resemble fictional narrative in their fairly
consistent use of the so-called 'epic preterite', i.e. the past forms charac-
teristically used in story-telling fiction.
316 Verbals
The fact that the loss of temporal deixis has given way to other, mainly
stylistic functions in narration has certain important repercussions for the
choice of progressive and nonprogressive forms (cf. Bache 1986: 9Iff). As
we have seen, we cannot normally use the simple present form of a verb with
dynamic meaning to instruct the hearer to identify a strictly present action
(for exceptions, see section 9.6.2):
(9a) *'He opens his packsack'.
(10a) *'He takes off his shoes and socks'.
Here only the progressive is possible. But in fiction the simple present is not
nearly as constrained. For the examples in (9a) and (lOa) to become
acceptable we only have to imagine a fictional context, such as a stage
direction or simply a piece of narration:
(9b) After a while David says 'Well children, time to break out the grass1. He
opens his packsack and gropes around inside.
(lOb) The two little girls go with him, then slip out of their shoes. Paul takes off his
shoes and socks, rolls up his trousers like an elderly tripper at the seaside.
Outside a fictional context, such simple forms have a more general habitual
or iterative meaning:
(9c) He always opens his packsack to check its contents.
(lOc) When going to bed, Paul usually takes off his shoes and socks.
9.7. Modal uses of tense-aspect forms

As shown in section 9.6.13, the English verb forms do not necessarily have
the same temporal meaning in indirect speech as in direct speech. The same
goes for tense-aspect forms when they are used modally. We return to
modality in section 9.9, but for the time being we can say that it centrally
involves non-factuality and concerns either degree οι probability (logical
possibility and necessity, hypothetical meaning, beliefs and predictability) or
desirability (permission, obligation, volition). In section 9.5 this was
illustrated by the modal verb MAY, which is used to judge the probability of
a situation in The economy may get worse and to express permission in May I
come in? The type of modality which concerns probability is termed
epistemic (derived from the Greek word for knowledge episteme) while that
which concerns desirability is termed deontic (derived from Greek deont-,
participial stem of del 'it is right').
In English all the tense-aspect forms except the simple present, the present
progressive and the present perfect can be used modally. In the following
account of these modal uses we shall disregard the use of the past forms in
conditional sentences, for that has already been dealt with, cf. section 8.9.
Modal uses of tense-aspect forms 317
Let us consider first the modal use of the past and the past perfect forms
(progressive as well as nonprogressive). As in conditional clauses these
forms can be used for the expression of hypothetical meaning, i.e. epistem-
ically, in object clauses and adverbial comparative clauses. This can be
illustrated by the following examples (which like most of the others in this
section are quoted from Davidsen-Nielsen 1990: 159ff):
(1) I wish I knew the answer to that.
(2) I wish I had known the answer to that.
(3) He talks as if he was writing a doctoral dissertation on the subject.
(4) He talks as if he had read the entire literature on this subject.
In (1) and (3) the hearer is not instructed to identify a past time situation but
is supposed to imagine a counterfactual situation in present time; and in (2)
and (4) he is not instructed to look back from a past time but is supposed to
imagine a counterfactual situation at a past time. What is common to this
modal use of a past form and its temporal use, however, is remoteness from
the world at the moment of speech, either in time or factuality (recall our
discussion of the past future in literary narrative in section 9.6.14).
Factual remoteness is also involved in those cases where the past is used
for the expression of (deontic) tentativeness and politeness:
(5) Could you do me a favour?
(6) We were wondering if we can expect to see you down here any time.
Here a situation is described as factually remote by means of a past form, and
this makes the request less direct - and therefore easier to turn down - than if
the present had been used.
In expressions like It's time we left ana It's time you went to bed, the past is
used deontically about that which is considered appropriate or necessary for
the subject to do.
Future (perfect) (progressive) forms can be used both epistemically and
deontically. The latter modal use can be exemplified by:
(7) You will do as I say at once.
(8) You will drop me by the cathedral in Leon.
(9) You will be studying hard in your room when I return.
Here the present future and the present future progressive do not instruct the
hearer simply to tag on to present time and then look ahead but express an
order. This modal use of the two forms is also closely related to their
temporal use. The order is presented in a non-negotiable way as a future fact,
and this makes it impolite and condescending.
318 Verbals
The future forms are used epistemically for the expression of nonfuture
predictability. The following examples express specific predictability:
(10) They will be home at this time of day.
(11) They will be watching telly now.
(12) It was 10 a. m. John would be in his office now.
(13) It was nearly midnight. The President would have written his letter of
resignation now.
Though modal tense-aspect usage is closely related to nonmodal usage here
as well, there is at the same time a clear difference. In using a sentence like
(10), for example, the speaker does not instruct the hearer to look ahead from
present time but informs him that a certain situation is predictable at the
moment of speech. In a sentence without a future time adverbial like John
will be in his office, the present future can be used either temporally to
instruct the hearer to look ahead or modally to express present time
predictability. Here the speaker can specify which meaning he has in mind
by adding a time adverbial, such as soon or now.
The future forms may also express habitual predictability:
(14) She will sit there for hours doing nothing.
(15) By four o'clock every afternoon he will have finished all his chores.
(16) Occasionally a tradesman's cart would rattle round the corner.
(17) Every afternoon between 4 and 6 he'// be preparing dinner.
Finally, the present future - but not the complex present future forms - may
be used to express general predictability:
(18) Oil will float on water.
(19) Sugar will dissolve in water.
Semantically, such present future sentences are closely related to sentences
in the present expressing a universal statement, like Water boils at 100
degrees centigrade (see section 9.6.2). Nevertheless, there is a difference in
meaning between e.g. Sugar will dissolve in water and Sugar dissolves in
water, for unlike the latter the former sentence contains an element of
aheadness that is a property of the temporal present future. This can be
captured by the paraphrase 'If sugar is put into water, it will dissolve*.
Modally used future forms are also found in the passive voice:
(20) This will certainly be rememberedby most people, (specific predictability)
(21) H2 SO4 will be revealed as an acid by the litmus test, (general predictability)
(22) Every morning she will be let in by the caretaker, (habitual predictability)
(23) You will be escorted by your mother as I told you. (order)
Mood 319
In allowing the passive voice, modally used future forms - like temporally
used future forms - differ from verb groups beginning with modal-volitional
WILL. This appears from the unacceptability of, for example, */ will gladly
be let in by the caretaker. There are therefore strong reasons for assuming
that the use of WILL-forms described in this section represents not regular
modal verb usage but secondary modal usage of future forms.
9.8. Mood
9.8.1. Introduction
Mood in English is traditionally regarded as an inflectional verbal category
with three members: the indicative (which has -s in the 3rd person singular of
the present), the imperative (realized by 0) and the subjunctive (also realized
by 0). These three moods can be illustrated by the following examples:
(1) Somebody opens the door (all the time).
(2) Somebody open the door (will you?).
(3) (I suggest that) somebody open the door.
The indicative and the imperative typically indicate what status the situation
referred to has, namely something which is real and something which needs
to be made real respectively (as illustrated by (1) and (2)). The communic-
ative functions statement and directive are thus typically implemented by
sentences whose verb stands in the indicative mood and the imperative mood
respectively (see section 4.4.3). The communicative function of the
subjunctive mood is less homogeneous, however. In an example like (3) it
appears to be directive like the imperative, in God save the Queen it
expresses a wish and in Lest anyone worry we're sinking, let me reassure you
we're not it describes a hypothetical situation. The subjunctive mood has
been termed 'thought-mood' by Jespersen, who terms the indicative 'fact-
mood' and the imperative 'will-mood' (1909-49, vol.7: 623). Though the
subjunctive differs from the indicative and the imperative in being
communicatively quite heterogeneous, the three moods are on a par
grammatically in forming a morphological paradigm.
As pointed out in section 9.7, modality centrally involves non-factuality, so
sentences in the indicative mood are here considered nonmodal (unless, of
course, they contain other modal expressions, such as She may join us later).
On the other hand, sentences in the imperative mood are modal (deontic) in
referring to situations which are not yet real; and sentences in the subjunctive
mood are non-factual and thus modal too (epistemic or deontic). Owing to its
factual nature the indicative is usually regarded as the unmarked, or neutral,
320 Verbals
member of the mood category. In the following discussion (which is based

on Davidsen-Nielsen 1990: 98-111), we examine the two non-factual moods.
Grammatically, mood differs from modality expressed by auxiliaries in
having its own paradigm and in being realized inflectionally, but semantic-
ally the two non-factual moods are closely related to the modal verbs. We
shall therefore analyse the imperative and the subjunctive by means of the
same concepts as the modals.
9.8.2. The imperative

In using the imperative the speaker typically issues a directive to the hearer
to behave in a specific way. Directives may be divided into a number of
different functional subtypes (see Huddleston 1984: 364f). The most
important of these can be illustrated by the following examples:
(1) Shut up. (command)
(2) Sit down, please, (request)
(3) Wash hair and rinse carefully, (instruction)
(4) Don't buy that brand if you can get others, (advice)
(5) Take as many sweets as you like, (permission)
(6) Give us this day our daily bread, (prayer)
Which of these subtypes of directive a sentence in the imperative mood
expresses is dependent on the linguistic and/or situational context. A marker
like please informs the hearer that an imperative probably functions as a
request, and if the speaker's relation to the hearer is that of a superior to his
subordinate, an imperative is likely to be understood as a command.
While the function of an imperative is typically directive, this is not
invariably the case. In sentences like Sleep well and Have a good time it is
used for the expression of a wish, and here it is clearly non-directive. Nor is
the imperative directive in a coordinated structure like the following:
(7) Do that again and I'll strangle you.
Here the speaker expresses his intention to inflict injury on the hearer if a
certain behaviour continues. The type of speech act performed is thus a
conditional threat (cf. If you do that again I'll strangle you).
As illustrated by the next example, a coordinated structure may also
express both a conditional threat and a directive:
(8) Shut up or I'll strangle you.
Here the speaker orders the hearer to stop behaving in a certain way and at
the same time expresses his intention to inflict injury on him in case this
directive is ignored (Unless you shut up I'll strangle you).
Mood 321
The imperative shows no present/past distinction (and is therefore perhaps

best considered nonfinite). Nor does it interact with the other choices in the
tense-aspect system apart from the fact that its combination with the
progressive cannot be ruled out entirely. But as it is not logical for the
speaker to direct the hearer to perform an ongoing action (e.g. IBe studying
and Ί Don't be studying), progressive imperatives are virtually never found.
The imperative is also highly restricted with respect to voice. This is not
surprising either, for in passive sentences the subject form never performs
the participant role DOER but typically the role DONE-TO (see section
7.4.1). Nevertheless, it is possible to issue a directive that the hearer should
avoid becoming the target (DONE-TO) of an action, and this is reflected in
the occurrence of negative passive imperatives like the following:
(9) Don't be taken in by that scoundrel.
While passive sentences of this type can be freely formed, an imperative like
Be seated cannot be regarded as a productive passive construction.
A special type of imperative which is very common has LET as its verb
followed by us (optionally contracted to let's):
(10) Let's go to the theatre.
(11) Let's not be petty-minded.
Here the directive is aimed not only at the hearer but simultaneously at the
speaker, as it is also in an example like Let you and me do the dishes.
Sometimes, however, we come across imperatives with LET which do not
express such a mutual proposal:
(12) Just let him try.
(13) Let AB be equal to B A.
Prohibitive imperatives are expressed by don't + V:
(14) Don't buy that brand if you can get others.
The auxiliary DO is also used in emphatic imperatives:
(15) Do help yourself to some more wine.
Even BE gets DO-support in such constructions, cf. Don't be late, Do be
careful. The negative of let's is let's not (e.g. Let's not be petty-minded),
though in BrE don't lets is possible as well (Don't let's waste more time).
When unaccompanied by us, LET requires DO-support (e.g. Don't let him
fool you).
Imperatives are typically subjectless (cf. section 5.2.2), but where a
contrast needs to be expressed, or emphatic displeasure, they contain the
subject form you:
322 Verbals
(16) You take the wine and I'll take the hamper.
(17) You mind your own business.
9.8.3. The subjunctive

Though the term 'thought-mood' would seem to indicate that the type of
modality expressed by the subjunctive is purely epistemic, the subjunctive
has not only epistemic but also deontic uses in English. What is common to
these uses is non-factuality, i.e. they both involve "something existing in the
speaker's mind only" (see Jespersen 1909-49, vol. 7: 624).
The epistemic subjunctive is used hypothetically in conditional, compar-
ative and concessive clauses. Outside formal language the only instance of
this is the use of were in combination with 1st or 3rd person singular subjects
(this verb form being standardly analysed as subjunctive). The occurrence of
this form in conditional clauses can be illustrated by:
(1) The task would be difficult if the old party were suddenly to reappear.
(2) Kinglake would have rung if the plane weren't on its way.
In BrE was is used in everyday language instead of subjunctive were in
conditional clauses, except in the fixed expression If I were you. Subjunctive
were is also found in the type of conditional clause signalled not by a
subordinating conjunction but by inversion (see section 8.9) and in compar-
ative and concessive clauses:
(3) The total output would be much worse were it not for the winter crop.
(4) It was as though no one else were there.
(5) Even if it were expedient I couldn't say what came before what.
In those cases where the situation described by a concessive clause is not
hypothetical but factual, was is naturally enough used instead of were, as in
Even if he was exceedingly well-off, I never considered marrying him.
In formal style, subjunctive were is used in fAaf-clauses (with or without
that) for the expression of wishful or hypothetical speculation (thus typically
in object clauses after the verbs WISH and SUPPOSE):
(6) I wish I were famous.
(7) Suppose a pretty high-school girl were to come forward with the solution.
In conditional and concessive clauses one occasionally also finds subjunctive
be. This usage is formal (cf. Quirk et al. 1985: 1093):
(8) If any vehicle be found parked on these premise, it shall be towed away at
the expense of the vehicle's owner.
(9) They did not approve of Atkinson who, be he a brewer in name, actually
committed the indignity of conducting trial marshings and fermentations.
Mood 323
Subjunctive be is sometimes also found in certain types of yes-no inter-

rogative and independent relative clauses:
(10) Whether a solution to this problem be found or not, we have to go ahead.
(11) Whatever be the reason, we cannot tolerate his disloyalty.
The subjunctive is used more frequently in AmE than in BrE. One example
illustrating this is its use after the conjunction lest:
(12) But lest it appear that I am always dishing it out, let me tell you, Miss Rose,
that I have often been on the receiving end, put down by virtuosi, by artists
greater than myself, in this line.
In BrE, lest is stylistically more marked (formal, old-fashioned) than in
AmE, like the subjunctive; and when it is used, it selects not only the
subjunctive but also should + V (e.g. Lest it should appear that...).
The deontic subjunctive is used for the expression of wishes in set
expressions like God save the Queen, Long live Trotsky and Heaven forbid.
The use of MAY for this purpose, for example in Please may it have been
instantly, will be dealt with in section 9.9.2.
The subjunctive is also used deontically for the expression of compulsion
in thai-clauses after verbs, adjectives or nouns expressing demand,
resolution, recommendation or the like:
(13)
demand
insist be fired
that Smith I leave at once
suggest
move
(14)
demand be fired
I support the suggestion that Smith leave at once
proposal
(15)
essential be fired
It is necessary that Smith leave at once
important
In AmE this use of the so-called mandative subjunctive is standard practice:

(16) Peg insisted that he see a psychiatrist in Providence.
(17) Christian ministers demanded that the studio destroy all copies of the film.
324 Verbals
In BrE, on the other hand, compulsion may also be expressed by should + V

in f/zaf-clauses, particularly in informal style (e.g. I suggest that Smith should
leave at once).
With verbs other than BE there is no formal way of deciding whether the
mandative subjunctive is used in an example of the following type:
(18) One of the campus ladies suggested that I urge him to shave his ears.
As the subject pronoun in the subclause is not in the 3rd person singular, the
absence of an -s inflection obviously does not demonstrate that urge is here
in the subjunctive mood. But the fact that urge is here synonymous with
should urge seems to indicate that this is the case, and so does the lack of
temporal agreement between urge and suggested. With subjects not realized
by 3rd person forms, lack of DO-support in negative f/za/-clauses constitutes
a non-inflectional criterion for identifying the mandative subjunctive: e.g. /
suggest that we not adopt this proposal.
A third deontic use of the subjunctive, which we shall characterize as
concessive, is found in frozen examples like these:
(19) Be that as it may, we'll still finish on time.
(20) Come what may, we'll proceed in the same way.
(21) Suffice it to say that he's totally incompetent.
In some cases - where paraphrases with LET are synonymous - the
concessive subjunctive comes close to expressing permission (compare e.g.
So be it with So let it be).
9.9. Modality
9.9.1. Introduction
By modality we understand a qualification of an utterance whereby the
speaker operates with alternatives to the actual world (see Davidsen-Nielsen
1990, on which this section is largely based, and from which examples are
extensively quoted). Human beings often think as if things might be other
than in point of fact they are, and for the expression of such conceptions they
use modal rather than categorical utterances.
As pointed out in section 9.7, modality primarily involves two kinds of
non-factuality: epistemic and deontic. In producing epistemically modalized
utterances like Perhaps Colonel Gaddafi is dead and Colonel Gaddafi may
be dead - which are qualified by perhaps and may respectively - the speaker
expresses that a certain situation is conceivably real. And in producing a
deontically modalized utterance like She ought to be in bed - qualified by
OUGHT - he expresses that a certain situation is desirable. With categorical
Modality 325
(non-modal) utterances like Colonel Gaddafi is dead and She is in bed, on

the other hand, the speaker describes situations which he considers factual.
In English, modality may be expressed lexically in a number of different
ways. This can be illustrated by examples like the following:
(1) Hopefully this is enough, (deontic adverb)
(2) She is likely to lose, (epistemic adjective)
(3) I permit you to smoke, (deontic verb)
(4) He's alleged to have resigned, (epistemic participle)
If modality is expressed grammatically, the qualifying element is either an
inflection - as in the case of the subjunctive and imperative moods, both of
which are realized by -0 (see section 9.8) - or a modal auxiliary, i.e. a form
of CAN, MAY, MUST, SHALL or WILL (see section 9.5). The expression of
modality by modal auxiliaries can be illustrated by epistemic examples like
the following:
(5) She may/can't/must be right.
Here the finite verb forms signal a qualification whereby the speaker
operates with alternatives to the actual world. And here the source of the
modality is the speaker of the utterance: Ί consider it possible/imposs-
ible/necessary that she is right'.
As Hoye (1997) observes, modality is often expressed by highly regular
collocations of modal verbs and certain adverbs:
(6) They can't possibly be playing tennis.
(7) It may well be a back-formation.
(8) It must certainly result in a termination of the project.
Though the modal verb and the adverb in such combinations express the
same kind of modality (e.g. epistemic modality), their combination is not a
matter of pleonastic reiteration of the same modal meaning but rather a
stylistically powerful, synergetic means of expression.
However, it is not always the case that modal auxiliaries are used for the
expression of modality. This can be illustrated by the next examples:
(9) Linford can run 100 yards in nine seconds.
(10) He can touch the ceiling.
Here CAN is used in factual statements about ability. As it does not signal a
qualification of an utterance whereby the speaker operates with alternatives
to the actual world, examples like these are nonmodal. As pointed out by
Palmer, CAN in the sense of ability "can be omitted from the strict
typological classification of modality, although it is of interest that modal
verbs have these meanings" (1986: 103). Notice also that CAN is here
326 Verbals
oriented not towards the speaker of the utterance (as in (5)) but towards the
referent of the subject form: the source of ability is the person referred to by
Linford in (9) and by He in (10).
In an important study on English modals by Klinge (1993), it has been
shown that the distinction between epistemic and deontic modality, i.e.
between expressions of probability and desirability, is not in fact a property
of the modal verbs themselves but of other elements in the sentence or even
outside it. For example, Jones may join us is likely to be understood as
deontic (permission) if followed by if he wants to but as epistemic
(possibility) if followed by as far as I'm informed. Whether Jones may join
us is understood as epistemic or deontic may also depend on the situational
context in which it is used. It thus seems that the distinction between
epistemic and deontic modality is not related to the modal verbs themselves,
i.e. is not a matter of lexical semantics, but is signalled by the context in
which they are used, i.e. is a matter of utterance pragmatics. When we say
that a modal is used epistemically or deontically, we thus imply that it occurs
in an utterance which is intended and understood to be either epistemic or
deontic. Similarly, a formulation like 'deontic CAN' is shorthand notation for
'CAN occurring in an utterance intended and understood to be deontic'.
As demonstrated in section 9.7, modality is closely related to tense-aspect
in that most of the tense-aspect forms can be used modally. The interaction
between these two categories also appears from the way the past forms of the
modal auxiliaries are used. As illustrated by the following examples these
forms are typically used not to instruct the hearer to tag on to world-before-
now but to express a weaker degree of modality than the present forms:
(11) She might be right.
(12) Could she be right?
These differ from She may be right and Can she be right? only in that the
degree of epistemic possibility expressed is relatively weak. Thus (11) can be
paraphrased as Ί don't rule out the possibility entirely that she is right'.
In the course of history the past forms of the English modals have in some
cases drifted away semantically from the corresponding present forms. This
goes for should, for example, which is used neither for the expression of
world-before-now nor as a weakened variant of shall. Examples illustrating
such semantic drift are You should be seeing them soon (epistemic) and
These guys should be jailed (deontic). In our discussion of the modals below,
we use small capitals for both formal variants (e.g. SHALL/SHOULD) to
stress their individual nature.
Modality 327
Finally we should mention that as they do not have nonfinite forms, the
modal auxiliaries often alternate with semi-auxiliary replacement forms (see
section 9.5). This can be illustrated by examples like the next ones:
(13) I can predict the result.
(14) Being able to predict the results is not enough.
(15) You must get up early.
(16) I hate having to get up early.
The expression of modality by means of tense-aspect forms (past, present
future, etc.) was discussed in section 9.7 and will not be taken up again.
9.9.2. MAY/MIGHT
In modalizing an utterance epistemically by means of MAY the speaker
indicates that a certain situation is conceivably real·.
(1) That may be the best light I'll ever appear in, to them.
(2) She may have felt possessive about Gertrude.
MAY is used deontically to express permission:
(3) May I speak to you for a moment?
(4) As far as forgiveness is concerned, you may look for it in this household.
In everyday informal language the use of MAY for the expression of a
personal permission is less frequent than that of CAN. An impersonal
permission is typically signalled by CAN, but occasionally it is expressed by
MAY. This usage is due to a prescriptive bias in favour of MAY (see Quirk et
al. 1985: 224):
(5) You may pay by direct debit, by post, or by bank giro credit.
Whether MAY is used epistemically or deontically sometimes has to be
determined on the basis of the extra-sentential or situational context (see
section 9.9.1). For example, a sentence like Jones may leave can mean either
that Jones's departure is conceivably real or that it is permitted. Note,
however, that in either case MAY is used for the expression of possibility:
presenting Jones's departure as conceivably real is the same as saying that it
is possible; and in granting Jones permission to leave, the speaker also makes
it possible for him to do so.
MAY (as well as MIGHT) is normally epistemic when followed by a perfect
and/or progressive form: Jones may be leaving /Jones may have left /Jones
may have been leaving.
Deontic MAY is sometimes followed by the adverbial (JUST) AS WELL:
(6) We may as well be straight with each other.
328 Verbals
(7) You may just as well hand in your resignation.

What the speaker typically indicates with this idiom is that a situation on
reflection has to be accepted even if this is not the instinctively preferred
reaction to it.
MAY is also used deontically for the expression of wishes and maledictions
(in sentences with partial inversion, see section 5.3.3):
(8) Please may it have been instantly.
(9) May it choke him.
In interrogative sentences, epistemic MAY is virtually ruled out. Here
epistemic possibility is expressed by CAN, as illustrated by Can it be true?
and Can spring be far behind? (which should be compared with corres-
ponding declarative sentences like It may be true and Spring may be far
behind). Deontic MAY, on the other hand, is used as readily in interrogative
sentences as in declarative sentences.
In negative sentences, the semantic scope of negation excludes epistemic
MAY but normally includes deontic MAY (see section 7.5.8):
(10) They may not have gone very far.
[Possible [not [they have gone very far]]]
(11) You may not borrow my car.
[Not [permitted [you borrow my car]]]
If the speaker wishes to describe a situation as 'not conceivably real' rather
than as "conceivably not real', he can do so by means of epistemic CAN
followed by NOT:
(12) They can't have gone very far.
[Not [possible [they have gone very far]]]
In sentences in the past, epistemic MIGHT is used to indicate that the speaker
is relatively uncertain about the possibility of the situation described by the
utterance. This use of MIGHT to convey a tentative attitude is illustrated by:
(13) That woman might attack you.
(14) What you suggest might be regarded as a recipe for folly and madness.
In examples like these, the past modal refers to non-past time and expresses
weak possibility (see section 9.7 on past forms and factual remoteness).
Epistemic MIGHT is also used without past time meaning in hypothetical
conditional sentences and in indirect speech (see sections 8.9 and 9.6.13):
(15) If I was a different person, it mightn't matter.
(16) She said the time might come in my generation when the educational system
itself was mixed.
Modality 329
Like epistemic MIGHT, deontic MIGHT is used with non-past meaning for
the expression of weakened modality. In choosing the past form the speaker
makes a permission tentative/hypothetical, and in interrogative sentences a
request for it will therefore be felt to be polite:
(17) Might I ask you to do me a favour?
MIGHT used as a tentative variant of deontic MAY is rare and is character-
istically restricted to questions and wishes (e.g. If only I might be allowed to
see him).
MIGHT is virtually never used to express past time permission. Exceptions
to this rule are found in subclauses of sentences with main verbs of thinking,
believing, feeling or knowing:
(18) She expected that I might come in about nine o'clock, but then our meeting
had to be cancelled.
9.9.3. CAN/COULD
Like MAY, CAN is used epistemically for the expression of what is
conceivably real and deontically for the expression of permission:
(1) Can spring be far behind?
(2) Can I stay out as long as I wish, Mum?
Let us consider first the epistemic use of CAN. As pointed out in section
9.9.2, CAN is used to express possibility in interrogative sentences:
(3) But can she be right?
(4) How can this be irrelevance?
(5) Whose beautiful antiques can these be?
Here CAN performs the same function as MAY in declarative sentences
(compare e.g. (3) with She may be right).
CAN is used also in negative declarative sentences:
(6) They can't have gone very far.
(7) You can't be serious.
What sentences of this type express is that the situation described is 'not
conceivably real', i.e. the semantic scope of NOT extends over the entire
sentence and thus includes the possibility modal.
In sentences in the past, epistemic COULD is used like MIGHT to convey
tentativeness, i.e. to indicate that the speaker is relatively uncertain about the
possibility of the situation described:
(8) He's not much here but he could arrive.
(9) But could she be right?
330 Verbals
In examples like these, the past modal refers to non-past time and expresses
weak possibility. Epistemic COULD is also used without past time meaning
in conditional sentences and in indirect speech:
(10) If we instructed him carefully, Jones could be the right man for the job.
(11) She asked me if it could be due to fear.
In positive declarative sentences, there are no restrictions on the epistemic
use of COULD, as illustrated by (8). On the other hand, epistemic CAN in the
sense of'conceivable that' (or 'possible that') - which characterizes its use in
interrogative and negative sentences - is ruled out in this sentence type. For
example, *She can have felt possessive about Gertrude is clearly ungram-
matical. Here the speaker has to use MAY instead.
Like MAY, CAN (as well as COULD) is normally epistemic when followed
by a perfect or progressive form: Can she be staying at the Park Lodge? /
They can't have left the hotel already / They could have left the stuff behind.
Before turning to CAN used deontically for the expression of permission,
we recapitulate that 'possible that* is expressed by CAN in interrogative
sentences and in negative sentences where the semantic scope of NOT
includes the modal (not-possible that) but that it is expressed by MAY in
positive declarative sentences and in negative sentences where the semantic
scope of NOT excludes the modal (possible-not that):
POSSIBLE THAT
Declarative Interrogative Negative

modal included modal excluded
MAY CAN CAN MAY
(12) She may be right.

(13) Can she be right?
(14) She can't be right.
(15) She may not be right.
If CAN is used deontically', the permission it expresses is often impersonal'.
(16) Make it clear that everyone can say anything.
(17) You can smoke in here.
In using a sentence like (17), the speaker often does not personally grant
permission to smoke but rather expresses permission indirectly by informing
Modality 331
the hearer that the rules allow smoking (cf. You can smoke in here, as far as
I'm informed). However, CAN is also widely used in informal style to signal
permission given by the speaker - or requested from the hearer - exclusively:
(18) All right, sweetheart. We're going. You can go back in now.
(19) Can I stay out as long as I wish, Mum?
Deontic CAN is sometimes used for the expression of compulsion:
(20) Toung man,11 say, 'you can just get up and leave this table. Leave the house
while you are at it.'
This example is clearly directive and produced by a speaker in authority. As
deontic CAN centrally serves the purpose of giving permission, the use of it
to give orders is rather condescending (like the deontic use of the future
form, see section 9.7). What it amounts to is approximately: You are
'permitted' to do something which I wish to see done. This directive effect, it
should be added, can also be obtained by MAY, for example in You may
leave immediately, young man.
The deontic use of CAN in negative sentences is illustrated by:
(21) You can't stay here.
Here - as in the case of epistemic can't - the semantic scope of NOT extends
over the entire sentence: [Not [permitted [you stay here]]].
In sentences in the past, deontic COULD is used with non-past meaning for
the expression of weakened modality:
(22) You could easily sleep here tonight.
(23) But couldn't we go to the cinema?
The past form makes a permission tentative, and in interrogative sentences
like (23) a request for it will therefore be felt to be polite.
Unlike MIGHT, COULD is freely used to express past time permission:
(24) In the late sixties we could do pretty much as we pleased.
In section 9.9.1, we discussed the nonmodal ability meaning of CAN in
positive declarative sentences like Linford can run 100 yards in nine
seconds. Consider now positive declarative sentences like the following:
(25) It can be cold in Stockholm.
(26) The exit can be blocked.
Here CAN is used in the sense of possible for. Note that these examples are
clearly different from 'possible that' examples like It may be cold in
Stockholm and The exit may be blocked. While the sentences with MAY are
synonymous with Perhaps it is/will be cold in Stockholm and Perhaps the
332 Verbals
exit is blocked, this is clearly not the case with the corresponding sentences
with CAN.
While sentences like (25) and (26) do not express ability directly - but
describe Ά property of the entity referred to (Stockholm, the exit) - it can be
argued that they express ability in a more abstract way. Note that in both
types of example the meaning of CAN can be captured by the formula
possible for: in the same way that it is possible for Linford to run 100 yards
in nine seconds, it is possible for Stockholm to be cold and for the exit to be
blocked.
CAN used in the sense of 'possible for' differs from epistemic MAY in
occurring freely in conditional sentences and in being ruled out before a
perfect infinitive with past time meaning or in progressive verb groups:
(27) If the exit can be blocked... (*If the exit may be blocked...)
(28) *It can be going either way. (It may be going either way.)
CAN used in sentences like (25) and (26) is thus not only semantically
similar to CAN of ability but also behaves like it syntactically (cf. If Linford
can run 100 yards in nine seconds .... *Linford can have run 100 yards in
nine seconds, * Linford can be running 100 yards in nine seconds). Therefore
it seems reasonable to group them together, and CAN is accordingly assumed
to be non-modal not only when it is used directly for the expression of ability
but also when it is used for the expression of what is a property of an entity.
In our view, non-factuality is not involved in (25) and (26), and CAN does
not in examples of this type constitute a qualification of an utterance
whereby the speaker operates with alternatives to the actual world.
In generic contexts the meaning 'possible for' can in formal English be
expressed not only by CAN but also by MAY (see Quirk et al. 1985: 223):
(29) Vampires can/may kill.
(30) Dogs can/may be snappish.
In examples of this type MAY is used non-modally, like CAN.
9.9.4. MUST
In modalizing an utterance epistemically by means of MUST, the speaker
indicates that a certain situation is necessarily real and that this is something
he infers from a set of facts. The modality involved here may therefore be
termed not only necessity but also deduction:
(1) I expect she hates me, why shouldn't she, she must be sore as hell.
(2) You must have made her think we'd been together.
If MUST is used deontically, it expresses compulsion:
Modality 333
(3) If the Labour Party disagrees with that assessment, it must give its reasons.
(4) Your friends will be going and you must go with them.
As pointed out in section 9.9.1, the distinction between epistemic and deontic
modality is a property not of the modal verbs themselves but is a matter of
utterance pragmatics. For example, Jones must be clever can mean either that
Jones's cleverness is inferred to be necessarily real or that it is compulsory
(e.g. required in a certain situation). Note, however, that MUST is used for
the expression of necessity not only when it is used epistemically but also
when it is used deontically: in compelling Jones to be clever the speaker also
makes it necessary for him to be so.
MUST is usually epistemic when followed by a perfect or progressive
form: She must be travelling with her brother again / They must have left
separately / He must have been doing the dishes. Strong contextualization
may, however, secure a deontic meaning of MUST in combination with the
progressive: You must be studying when I return.
Let us examine first the epistemic use of MUST in different sentence types.
As illustrated by the next example we find it in wh-interrogative sentences:
(5) What must it have been like in the Middle Ages, I wonder?
In yes-no interrogative sentences, on the other hand, MUST is very rare and
restricted to those cases where a negative orientation has to be conveyed. For
example, a speaker may p*se a question like Must he be on holiday? if he
challenges a claim that the person referred to is on holiday. Otherwise,
epistemic necessity is expressed by the semi-auxiliary NEED in this sentence
type (see section 9.5):
(6) Need this suggestion have any party political implications?
(7) Need it have happened that way?
In negative declarative sentences, epistemic MUST is not used. Here
necessity is expressed by the semi-auxiliary NEED:
(8) It needn't affect the incidence of local taxation.
(9) Survival after death and unending improvement need not mean perfect
happiness.
What sentences of this type express is that the situation described is 'not
necessarily real', i.e. the semantic scope of NOT extends over the entire
sentence and thus includes the necessity modal: [Not [necessary [it affect the
incidence of local taxation]]].
The expression of epistemic necessity in positive declarative, interrogative
and negative declarative sentences can be summarized and exemplified like
this:
334 Verbals
EPISTEMIC NECESSITY
Declarative Interrogative Negative

yes-no wh- modal included modal excluded
MUST NEED MUST NEED
(10) Boris must have committed this crime.

(11) Need Boris have committed this crime?
(12) Who must have committed this crime?
(13) Boris needn't have committed this crime.
As appears, there is in English no negated modal form available for the
expression of necessary-not. However, the speaker may obtain a comparable
semantic effect by selecting the possibility modal can't. The meaning of a
sentence like Boris can't have committed this crime can be captured by the
notation [Not [possible [Boris commit this crime]]], and this is logically the
equivalent of [Necessary [not [Boris commit this crime]]]. Note in this
connection that example (13) is semantically close to Boris may not have
committed this crime. This is due to the fact that [Not [necessary [Boris
commit this crime]]] is logically the equivalent of [Possible [not [Boris
commit this crime]]]. In this case, however, the speaker has a choice, and
Boris needn't have committed this crime and Boris may not have committed
this crime are not entirely synonymous, for there is a difference between
making a judgement in terms of necessity and making it in terms of
possibility. That epistemic MAY NOT and NEED NOT are not always
interchangeable is apparent from an example like the following (see Palmer
1990: 61):
(14) He may be there, but he needn't be.
In sentences in the past, weakened epistemic necessity can be expressed by
SHOULD, as in They should have reached their destination by now. We
return to this use of SHOULD in section 9.9.5.
Let us now turn to the deontic use of MUST. As pointed out above,
sentences with deontic MUST express compulsion, and they are often used as
directives (see e.g. example (4) above). In sentences where the subject is in
the first person, compulsion is typically directed towards the speaker himself,
either exclusively (singular) or inclusively (plural). In this way the speaker
appeals to himself, i.e. I/we must has the meaning of self-admonishment (see
Quirk et al. 1985: 225):
Modality 335
(15) I must tell them that some other time, it's a separate story.
(16) We must go round to Tim's place, at once, all of us.
The deontic use of MUST in interrogative sentences can be illustrated by the
following examples:
(17) Must I go back to school so soon?
(18) But why must I sit here?
In yes-no interrogative sentences like (17), the speaker asks the hearer to
decide a course of action for him. In wA-interrogative sentences like (18) the
hearer is asked to explain the particular reason, place, time or identity of a
directive. In some interrogative sentences with MUST where the subject is in
the second person, the speaker indicates annoyance with the hearer:
(19) Must you discuss all the time?
(20) Why must you be so stubborn?
In interrogative sentences, compulsion may also be expressed by sentences
with the semi-auxiliary NEED. In choosing NEED - which is rare in wA-
questions except those beginning with WHY - the speaker also asks for a
directive but at the same time asks the hearer whether the course of action
described by the interrogative sentence is strictly necessary. NEED is
therefore often accompanied by the adverbial at all:
(21) Need she participate in the meeting at all?
(22) Why need I stay at home at all?
In some interrogative sentences, NEED differs from MUST in that the latter
implies that the hearer has the power personally to decide a course of action
(see Palmer 1990: 78):
(23) Need I stay at home tonight?
(24) Must I stay at home tonight?
The deontic use of MUST in negative declarative sentences can be illustrated
by examples like the next ones:
(25) You mustn't think that I don't understand your feelings, my dear.
(26) You mustn't take me for an old fool with his head in the clouds.
In sentences of this type, the semantic scope of negation excludes the modal:
[Compulsory [not [you take me for an old fool]]]. If the speaker wishes to
describe a situation that is not-compulsory, i.e. in such a way that the
semantic scope of negation includes the modal, he must select the semi-
auxiliary NEED:
(27) And you needn't glare at me like that.
336 Verbals
(28) You needn't high-hat me.

The semantic contrast between deontic MUST and NEED in negative sen-
tences can be illustrated by examples like You ^mustn't reply (compulsory-
not) and You needn't reply (not-compulsory).
MUST NOT is also used with first person subjects for the expression of
self-admonishment:
(29) Anyway, I mustn't anticipate.
As compulsory-not is closely related to not-permitted, replacement of MUST
by CAN or MAY would not result in any significant change of meaning in
negative sentences, cf. an example like You may not/can't (mustn't) park
here. As pointed out earlier in this section, however, there is a certain
difference between making a statement in terms of possibility (permission)
and making it in terms of necessity (compulsion).
In sentences in the past, weakened compulsion can be expressed by
SHOULD, as in These guys should be jailed. We return to this use of
SHOULD in section 9.9.5.
A frequent alternative to MUST for the expression of epistemic necessity or
deontic compulsion is the semi-auxiliary HAVE TO (see section 9.5):
(30) There must/has to be a way out. (epistemic)
(31) You must/have to do it at once, (deontic)
9.9.5. SHALL/SHOULD
Though SHALL and SHOULD are historically different forms of the same
lexical item, they are used for very different purposes, and we shall therefore
discuss them separately.
Compared with the other modals SHALL is very rare, particularly in AmE.
As illustrated by the following examples it is used deontically for the
expression of commitment in declarative sentences with subjects in the
second or third person (SHALL with first person subjects being a formal
substitute for the future tense form WILL, see section 9.6.3):
(1) You shall have your car back by Friday.
(2) Our children shan't ever bother you again.
(3) It shall be delivered tomorrow, sir.
According to Quirk et al. (1985: 230) this use of SHALL is archaic, and in by
far the majority of cases the natural choice of modal here is WILL. Although
this verb does not commit the speaker to the same degree as SHALL, the two
verbs are nevertheless semantically close enough for the replacement of the
latter by the former to be understandable.
Modality 337
SHALL is also used deontically for the expression of (relatively weak)

compulsion, though only in interrogative sentences with first person
subjects:
(4) Shall we go to the theatre?
(5) Where shall I put it?
The meaning signalled by Shall I/we is here 'Do you want me/us to ...?' As
pointed out by Quirk et al. (1985: 231) this meaning is more commonly
expressed by Would you like (me/us) to or Should I/we, particularly in AmE.
In declarative sentences, the use of SHALL for deontic compulsion is only
found with third person subjects to denote what is legally mandatory, i.e. is
restricted to ESP (see section 1.8):
(6) It shall be unlawful to carry firearms.
(7) The tenant shall quietly possess and enjoy the premises during the tenancy
without any interruption from the Landlord.
SHALL is never used epistemically.
Let us now turn to the past form SHOULD. When this modal is used
epistemically, the speaker expresses that he expects the situation described to
be real but does not feel absolutely certain, for example because the facts
upon which his deduction is made may not be complete. The use of SHOULD
to indicate such tentative certainty can be illustrated by examples like:
(8) They should have reached their destination by now.
(9) You should be seeing my family tomorrow.
In sentences like these, SHOULD refers to non-past time (in fact, SHOULD
never refers to past time) and expresses weakened necessity, which is the
equivalent of high but not maximum probability.
The epistemic use of SHOULD in negative sentences is illustrated by:
(10) It shouldn't be difficult to accomplish that.
Here the semantic scope of NOT excludes the necessity modal: [Weakly
necessary [not [it be difficult to accomplish that]]].
The fact that SHOULD indicates tentative inference readily explains its
common occurrence in conditional sentences:
(11) If I remember right, this region should be just above the great face.
(12) If you press that button, the engine should start.
Epistemic SHOULD is also common in the subclause (with inversion) of
conditional sentences (see section 5.3.3):
(13) Should you happen to be passing, do drop in.
338 Verbals
A special use of epistemic SHOULD is found in examples like:

(14) Well, it surprises me that Eileen should be surprised.
(15) I can't think why he should have been angry.
According to Palmer, who quotes the first of these examples (1990: 189), the
modal is here semantically empty, whereas Quirk et al., who cite the second
(1985: 234) are of the opinion that it expresses 'putative' meaning. It might
also be argued, though, that SHOULD is here used for the expression of
report, i.e. to signal that the speaker passes on something he has heard about
and the truth of which he is therefore not committed to. This use of SHOULD
may be compared with the quotative use of German SOLLEN in e.g. Er soll
reich sein 'He is said to be rich'.
Deontic SHOULD is used with non-past meaning for the expression of
weakened compulsion. By this choice of modal a directive is made tentative
and therefore less insistent. What is indicated in this way is often no more
than a suggestion:
(16) These guys should be jailed.
(17) Perhaps we should draw a veil over the last year or so.
(18) Etiquette demands that I should invite him.
As illustrated by (18), one context in which SHOULD of compulsion occurs
is /Aai-clauses after verbs, adjectives or nouns expressing demand,
resolution, recommendation and the like (see the discussion in section 9.8.3
of the mandative subjunctive, which also occurs in this context).
The deontic use of SHOULD in negative sentences is illustrated by:
(19) It shouldn't be supposed that Stone identifies with Socrates in any overt way.
Here the semantic scope of NOT excludes the modal: [Weakly compulsory
[not [Stone identify with Socrates in any overt way]]].
Apart from the fact that it expresses compulsion tentatively, deontic
SHOULD often differs from deontic MUST in indicating that the situation
described is morally desirable:
(20) A man shouldn't leave his home. Or he'll become a wanderer, a lost soul.
(21) Fictional characters should, whenever possible, reflect the non-smoking
majority of the population.
The meaning expressed by SHOULD - whether epistemic or deontic - can
also be expressed by the semi-auxiliary OUGHT TO (see section 9.5):
(22) They should/ought to have arrived by now.
(23) We should/ought to have declared war on Hitler when he marched into the
Rhineland.
Modality 339
Though SHOULD and OUGHT TO are not invariably interchangeable, and

have different patterns of stylistic variation, they are normally synonymous.
9.9.6. WILL/WOULD
As the use of purely temporal WILL/WOULD in future forms has already
been dealt with (in section 9.6.3), it is only the use of volitional
WILL/WOULD which needs to be accounted for here. The use of the present
form can be illustrated by examples like the following:
(1) I think today I'll stick to cheese.
(2) I will never leave you, father.
(3) Will you come with me, Sancho?
(4) If she'll wait in the study, I can see her in a minute.
Here WILL expresses volition, but at the same time it has futurity as a
constant secondary meaning. In this way it differs from MAY, CAN and
MUST (though not SHALL), for in some of their uses these modals do not
describe future situations, cf. e.g. This coat may be John's, How can this be
irrelevance? and You must be tired.
Like CAN of ability, volitional WILL is neither used epistemically nor
deontically but in factual statements about willingness or intention. As it
does not signal a qualification of an utterance whereby the speaker operates
with alternatives to the actual world, the meaning of examples like (1)
through (4) is non-modal. Like possible-for CAN, furthermore, volitional
WILL is oriented not towards the speaker but towards the referent of the
subject form. In an example like Will you post this letter for me? the source
of volition is thus the person referred to by you, not the speaker of the
utterance.
In negative sentences, the semantic scope of NOT usually includes WILL.
For example, a sentence like / won't put up with his behaviour any longer
clearly indicates that the person referred to by / is unwilling to accept the
situation described. Sometimes, however, WILL is excluded from the
semantic scope of negation. This is the case in We won't bother you any
more, which indicates that the persons referred to by we are willing not to be
a nuisance any longer.
Not surprisingly, volitional WILL is ruled out in passive and progressive
sentences (see section 9.6.8 on temporal WILL + B E + V-ing) and before
perfect infinitives. The finite auxiliary occurring in e.g. You won't be
bothered, They'll be leaving in half an hour and The committee will have
finished its work soon is thus not volitional but purely temporal.
340 Verbals
WILL is occasionally used for what has been termed strong volition (see
Leech 1987: 86):
(5) He 'will go swimming in dangerous waters.
When used for this purpose, i.e. in the sense of 'insist on1, WILL is
obligatorily stressed and cannot be contracted to ΊΙ.
The past form WOULD is used with non-past meaning for the expression of
•weakened volition, often in polite requests:
(6) Would you pay us in cash, please?
(7) You wouldn't have the time to do it now, would you?
(8) We would like to sit down soon.
Volitional WOULD is also used without past time meaning in hypothetical
conditional sentences and in indirect speech (see sections 8.9 and 9.6.13):
(9) If you did that, I would bash in your brains.
(10) My wife said she would phone us after dinner.
Here it seems clear that WOULD serves the purpose of expressing not only
aheadness but also intention.
10. Nominals
While verbals typically express situations (cf. sections 7.2 and 9.1), nominals
typically express the participants involved in situations, e.g. the agent, the
affected or the instrument. The main communicative function of nominals is
thus to code meaning as things (or 'entities') in a broad sense (concrete as
well as abstract, animate as well as inanimate). This function is very
composite, involving many different lower-level communicative functions,
such as determination and modification. It can therefore be thought of in
terms of a functional domain, by which we mean a general main function
comprising a number of subfunctions. In this chapter we shall examine the
ways in which nominals occupy this functional domain, i.e. how they enable
the speaker to 'talk about things'.
10.1. Preliminaries
10.1.1. Nouns and noun groups
The present chapter deals with both types of nominal: single nouns and noun
groups. As will be recalled, nouns constitute a major word class comprising
items which typically express things (e.g. BOOK), and which are often
combined with articles (e.g. the book) and inflected for the expression of
number (e.g. book/books) and the genitive case (book's) (cf. section 3.1.4). A
noun group is defined as a group with a noun as head (cf. section 3.3.1):
(1) This must be familiar scenery.
(2) Moira is smoking a cigarette. She takes a drag, passes it to her husband.
(3) [This time} [her eyes] give him a penetrating stare.
However, we have to interpret this definition broadly to accommodate more
complex cases involving stacked heads larger than the noun, as in the fol-
lowing examples (see section 4.1.1 on such stacks):
(4) We were served the big brimming jugs of cream you only see at farming
functions.
(5) He was infinitely more interesting than comparable apocalyptic zealots who
were characteristic of that period of Jewish history.
In a stack analysis, the nouns jugs and zealots are assigned head status, not at
the primary group level but at a much lower constituent level.
10.1.2. The external relations of nominals

Nominals may assume the following functions:
342 Nominals
S The restaurant was crowded.

Od I was drinking vintage champagne.
Oi I finally told Jack's wife my little secret.
Cs Most of the diners were tourists.
Co We elected Irene our first female director.
A This time the bastards won't get away with it.
In addition, nominals may serve as dependents (e.g. in noun, preposition and
adjective groups) and as conjoints in compound units:
DEP These solar energy schemes were proposed by my boss.
The figure was ten inches tall.
CJT The organist and the photographer were hired.
Note also that nominals may serve a number of communicative functions
directly, e.g. EXC (Christ!) and QUE (A book?).
10.1.3. The internal structure of noun groups

By definition, nouns assume head function in noun groups. Dependents
appear in either pre-head or post-head position. We can thus offer the
following first approximation to the structural potential of noun groups:
pre-H dependents H post-Η dependents
As pre-H dependents we typically find:

(i) articles (as in a_ bed/an honour/fhe boat)',
(ii) possessive pronouns and nominals in the genitive (as in her book / my
speech /John's pen / the old professor's office /my sister's new book)',
(iii) demonstrative, interrogative and relative pronouns (as in this girl /those
plays / which book/ whose idea /what students);
(iv) indefinite pronouns and quantifiers (as in some sugar /any woman /no
entry /every word / neither statement / many proposals /all the letters);
(v) adjectivals (as in excellent teachers / solar energy / a very interesting
idea /a most original exhibition);
(vi) present and past participles (as in a dancing girl / rising prices / returned
goods / a defeated enemy);
(vii) nominals (as in a development plan /university students / a civil rights
movement /a fifth Middle East war).
Preliminaries 343
Occasionally we find the following as pre-H dependents:

(viii) adverbs (as in the then king / the above examples / the in thing)',
(ix) complex group- and clause-like structures, the unity of which is usually
marked by means of hyphenation (as in a two-year-old boy / the latest
Humphrey-for-President movement/a small, what-else-can-you-expect nod).
As can be seen, the relationships between pre-head dependent and head are
typically determination (as in a bed / her book) and modification (as in
excellent teachers / university students).
In post-head position we typically find:
(i) preposition groups (as in a letter from my uncle / a rule of this kind / a
town in Germany / a visit to my parents}:
(ii) relative clauses (as in the letter which you wrote last night /John, who
moved to Hove last year. / the film that you found so interesting);
(Hi) single or coordinated adjectivals (as in the only stars visible /professors
keen to take early retirement / the leaves, so soft and yellow):
(iv) nominals (as in the meeting last night / the match next week / Jack
Parker, my neighbour. / our new manager, the tall guy who just left. / the
number six / my dear friend Richard}:
(v) present and past participles and participial clauses (as in all the guests
leaving / the prisoners deported / the colleagues remaining behind / some of
the cars tested last month):
(vi) infinitive groups or infinitive clauses (as in any attempt to move / the
decision to break up the party}:
(vii) non-relative fAaf-clauses (as in the fact that she wants to leave /the idea
that I should marry her / the hope that someone will step in and rescue her):
(viii) adverbals (as in the meeting inside / the book here / the road back).
As these examples show, the most common relationships between head and
post-head dependent are modification (as in a letter from my uncle) and
complementation (as in any attempt to move), cf. section 6.3.3. But we also
get what may be referred to as 'elaboratives', i.e. dependents which enter an
identity relation to the head but at the same time elaborate on the content of
the head: Jack Parker, my neighbour, —» 'Jack Parker = my neighbour', the
number six —> 'the number = six'. In some constructions with clausal
elaboratives, the head noun serves primarily as a means to nominalize the
content of the dependent clause: the fact that she wants to leave —» 'the fact =
she wants to leave', the idea that I should marry her —» 'the idea = I should
344 Nominals
marry her'. In such constructions the head noun is always abstract (other
examples: ANSWER, BELIEF, CHANCE, CLAIM, NEWS, POSSIBILITY,
PROPOSAL, SUGGESTION).
Post-head parenthetical dependents, i.e. dependents which are separated
from the head by means of intonation or commas, are often referred to as
appositional (e.g. Jack Parker, my neighbour. /John, who moved to Hove
last year.}. Some grammarians use the term 'apposition' to refer to elabora-
tives, whether parenthetical or not.
There are in general fewer post-head dependents in noun groups than pre-
head dependents. On the other hand, post-head dependents are often realized
by preposition groups or by clauses and thus tend to be longer than pre-head
dependents, which are often realized by articles, pronouns and adjectives.
While many post-head dependents are thus fairly complex, many pre-head
dependents are fairly simple (being realized by single words or two- or three-
word groups) but form more complex relationships with the head noun and
each other. In order to describe these facts adequately it is important to relate
the fairly automatic division of the noun group into pre-head, head and post-
head to the functional domain of the noun group and to the fragmentation of
this functional domain into subfunctions.
10.1.4. The functional domain of nominals

As mentioned in the introduction, the functional domain of nominals can be
defined in traditional terms as the expression of meaning as 'things'.
Nominals enable us to code what we want to talk about as things with the
degree of specificity required for our communicative purposes: speakers
encode meaning in nominals in the shape of things and listeners decode such
constituents accordingly. Nominals are used for a variety of more specific
communicative functions, such as to identify specific things:
(1) The restaurant was crowded.
(2) The bastards won't get away with it.
or to mention 'type ofthing':
(3) I told her I wanted an apple.
(4) Teachers who work overtime must be very idealistic.
or to describe an already identified thing:
(5) Most of the diners were Japanese tourists.
(6) Rose is a very good student.
or to specify the 'situation' expressed by the verb in examples of syntactic and
semantic fusion between predicator and object (cf. section 7.3.4 [I]):
Preliminaries 345
(7) The meeting took place yesterday, (took place = 'happened')

(8) We caught sight of her. (caught sight of= 'saw1, 'sighted')
In all these examples the italicized nominals code meaning as things, but the
things coded are used for different purposes. To get a better understanding of
the relationship between nominals and the things expressed, it is useful to
examine examples like the following:
(9) He found the small yellow key in the kitchen sink.
(10) A glass of orange juice is just what I need now.
In the first example, the two noun groups the small yellow key and the
kitchen sink can be understood to 'refer to1 things in 'the real world' or in
some 'fictional world' (e.g. in a novel), i.e. they identify a particular key and
a particular sink, respectively. In such cases we can say that there is a
relation between language and specific 'extra-linguistic' things, a relation in
terms of reference. In the second example, the noun group A glass of orange
juice can hardly be said to refer to a specific thing in this sense but to
mention something that could become a specific thing of the world. For
many linguists, 'reference* in connection with a nominal means 'reference to a
specific thing'. Thus the small yellow key (as well as the kitchen sink) is
usually regarded as a referring expression, whereas A glass of orange juice is
not. In this grammar, however, we use the term 'reference' in a broader sense
to include any link between a formal expression and our conception of the
things expressed. There is thus a strong conceptual element in all reference,
as we use the term. When we 'refer to' things, general or specific, we do so
on the basis of how we conceive of them rather than on the basis of their
objective reality. This is particularly clear when abstract 'things' (such as
ideas, relations, conventions, etc.) and non-existent 'things' are involved
(such as unicorns). Note also examples like (7) and (8) above, in which
something that is not objectively 'a thing' is represented as 'a thing'. But even
reference to, say, a key or a glass of orange juice has a conceptual, cognitive
basis in the sense that we must share an understanding of 'key' and 'glass of
orange juice1 to be able to communicate about them. This understanding of a
thing can be described as a mental representation. We use 'reference' about
the link between expression and mental representation rather than the link
between expression and things in the 'real world'.
The capacity of nominal reference (broadly defined as our capacity to
express meaning in terms of things) is dependent on a number of com-
municative subfunctions. In the following, we shall take a closer look at
these subfunctions.
Nominal reference normally relies on cognitively 'fat' lexical words (such
as the head noun) as well as cognitively 'lean' function words (such as
346 Nominate
determiners), both contributing to a set of contrasts, some of which are

context-dependent, others more context-independent. To describe the
resulting delimitation (or 'singling out') of things, we use the term 'contrast-
formation': nominate provide a set of contrasts with which to discriminate
between things. Roughly, contrast-formation seems to work as follows. By
using the definite article in a nominal like the small yellow key, the speaker
signals that there will be enough information in the nominal for the listener
to construe the referent (i.e. the thing expressed as we conceive it) as unique
in the context. This functionally derived contrast between unique and non-
unique is combined with three lexical contrasts, the first two of which are
provided by the adjectives small and yellow. The meanings denoted by these
('smallness' and 'yellowness') must be considered in relation to the third and
more context-independent contrast, the one provided by the head noun key.
The particular combination of various contrasts evoked by the small yellow
key enables the listener to construe the meaning intended by the speaker.
There is often more contrast-formation in nominals than we actually need
in order to identify the intended referent in a particular context. In e.g.:
(11) Have you met my beautiful wife yet?
we cannot say that the adjective beautiful is necessary to establish who the
referent of the construction my beautiful wife is, unless of course the speaker
has more than one wife and only one of them is beautiful. Instead it simply
offers a description of the referent. Though this description clearly involves a
(subjective) contrast (beautiful vs. non-beautiful), this contrast is, strictly
speaking, redundant from the point of view of establishing the referent: an
expression like my wife will usually do the job with sufficient precision. In
general, whether or not contrast-formation is redundant may depend entirely
on the context of the utterance. Consider:
(12) The unhappy mother left at once.
If the context is such that there is only one mother present, the adjective
unhappy is 'merely' intended as a description of that mother: it could be left
out without making the noun group referentially unclear. If the context is
such that there are several mothers present but only one of them is unhappy,
unhappy provides information without which the listener cannot establish
which mother left: without the adjective the noun group becomes refer-
entially unclear. For the listener, the communicative status of the contrast
provided by unhappy as either descriptive elaboration or identificatory clue
must be worked out in context.
We may accordingly distinguish between restrictive and non-restrictive
contrast-formation. This distinction is a fundamental, but also very general,
Preliminaries 347
subftmction of the functional domain of nominals. The following chart

provides a first approximation to our description of the relationship between
communicative function (shaded cells) and internal structure (white cells):
expression of meaning as things
contrast-formation: it»tri<îve/iioil-i5es*rictive
''""""' · · '
pre-H H post-H
One of the factors contributing to the pre-head complexity mentioned in

section 10.1.3 above is the presence of two regular communicative sub-
functions: determination and modification (cf. section 6.3.3). Determination
and modification, as realized by pre-head dependents, can be thought of in
terms of zones in the noun group arranged more specifically in the following
order:
determination modification H post-H
Determination is realized by articles, pronouns and genitive constructions,

while modification is chiefly realized by adjectivals:
. . . . . . - : - . . . . ' ·
determination modification Η post-H
:;
Ϊ:· me ;;: little girl with the shy smile
an old friend from London
mis very dull visit to her parents
no additional staffing, academic or secretarial,
my best student, who left school early,
j the sudden death of my father
my neighbour's thick fibrous clothes from Woolworth's
The communicative function performed by the head of a noun group is to

provide a close lexical match for the referent of the construction. The head
thus represents the referent as a member of a category of the things, persons,
348 Nominate
etc. In e.g. the little girl with the shy smile, the head noun girl categorizes the
person referred to as a girl (rather than as e.g. a woman, a boy or a man). The
functional nucleus of the noun group is thus categorization:
determination modification ^ ^ post-H
Post-head dependents are used for a variety of communicative functions:

determination, modification, categorization and complementation (see
section 6.3.3). Thus, for example, in the sudden death of my father, the post-
head dependent of my father is determinative in conjunction with the definite
article (cf. my father's sudden death, where my father's is pre-head
determination). In the little girl -with the shy smile, the post-head preposition
group with the shy smile is clearly a modifier on a par with little, describing
the head noun girl. In no additional staffing, academic or secretarial, the
post-head compound unit academic or secretarial offers a subcategorization
of the head noun staff. And finally, in this very dull visit to her parents, the
post-head preposition group to her parents serves as complementation to the
head noun visit. These many post-head functions do not often co-occur and
therefore do not represent as well-established zones as the pre-head
functions. Accordingly we shall simply operate with one post-head multi-
ßinctional zone:
- - - -· - - ; ' -.·· " . ' ' .. " -· · ·· · .-- ··. . .' . ·-·;··. . ·- ..:.
determination
- - ~ :- : .. - ;.
.
...:.·.
modification .
categorization
.· " . - . - . · - " :
tailti-fonctionaf)
. " ·-··."·
One important communicative function - related mainly, but not exclusively,

to the head and pre-head constituents - is quantification. Given the nature of
the functional domain of nominals, it is natural for quantification to play a
central role: when we talk about things it is often essential to signal their
quantity. In any discussion of quantification, it is important to draw a
distinction between countable and non-countable things. Countable things
are things that we think of as something we can count: cars, houses, books,
records, etc. Non-countable things are things that we think of as masses of
some sort and which we do not usually count (though there are other ways of
measuring them): water, flour, sand, sugar, etc. Countable things are
indivisible while non-countable things are divisible in the following sense:
we can divide water into parts, each part still being water, but we cannot
divide a car into parts and regard each part as a car by itself. The
singular/plural distinction in nouns referring to countable entities (e.g.
boy/boys, girl/girls, man/men, woman/women, etc.) is at the very heart of
Preliminaries 349
quantification but by no means the only way of expressing this commun-

icative subfunction, which also involves non-countable concepts: we can talk
about more or less water, flour, sand, sugar, etc. Here are some examples:
(13) These ten books are far too expensive.
(14) A short meeting took place last night.
(15) She complained about my numerous girlfriends.
(16) Some students seem to think that life is a bowl of cherries.
(17) It took little effort to finish the job.
(18) Students in great numbers have cancelled their participation.
As these examples show, quantification is often expressed in connection with
determination (These, A, Some) and categorization (singular meeting and
plural books, girlfriends, students) but sometimes also in connection with
modification (short, numerous, ten, little, in great numbers).
The following chart summarizes our discussion so far of the structure of
the nominal in relation to its functional domain:
--- ;-- . ;·
' '- .V ·;·- ' expression of nicanint? as thincs
contrast-formation: restrictive/non-restrictive
.' . Λ · · . . ,.·· ;. !§;/;S|7:: :iif ···: quantification \
determination modification categorization (multi-functional)
To conclude, as this chart shows, nominals are used to express meaning in

the shape of entities. The primary, but also most general means of serving
this functional domain is contrast-formation (restrictive/non-restrictive). The
functional domain of expressing meaning as entities is fragmented into
contributing subfunctions, each more or less distinctly associated with the
zones created within the pre-head, head and post-head structure of the noun
group. The expression of entities often requires an interplay of two or more
of these subfunctions. The rest of this chapter is devoted to a discussion of
categorization (in section 10.2), determination (in section 10.3) and
quantification (in section 10.4). Modification will be examined more closely
in chapter 12 on adjectivals.
350 Nominals
10.2. Categorization
10.2.1. What's in a head?
The central categorizing unit in the noun phrase is the head. The head
typically consists of just a noun:
(1) Sometimes the boys could entice the old janitor to time them with the official
clock.
(2) Reagan is proud of every effort he's made for the contras.
However, as pointed out in section 2.1, there are cases where two or more
root forms that function independently in other circumstances seem to
constitute a single lexical item as head, a compound:
(3) We are now approaching the airport.
(4) She simply adored her mother-in-law.
(5) Have you met our new dancing master!
(6) He takes a professional interest in the human nervous system.
In writing, the unitary status of these items is sometimes indicated by
absence of an empty space or by hyphenation, as in (3) and (4). In speech,
compounds consisting of two elements typically take main stress on their
first element, i.e. they are pronounced with so-called unitary stress. Thus
'dancing jnaster (= master of dancing) contrasts with the syntactic group
tdancing 'master (= master engaged in dancing), and the first element of
'nervous ,system contrasts with the first element in the group tnervous 'girls.
Sometimes, as in French teacher, stress is criterial for our classification of a
word as either a noun or an adjective: with unitary stress on the first element
('French Jeacher), French is a noun forming a part of a compound with the
meaning 'teacher of French'; with main stress on the second element (,French
'teacher), it is a premodifying adjective denoting the nationality of the
referent of the head noun. Stress is sometimes a difficult criterion to work
with, however. For example, some well established combinations like
headmaster and ginger ale take main stress on their second element like
syntactic groups such as head waiter and ginger hair and only differ
prosodically from these groups in that their first element tends to be more
weakly stressed. Our policy is to keep the head as simple as possible and
only allow clear cases of compounding.
There are cases, however, where it is convenient to treat orthographically
complex units as heads because they are fixed collocations, often resisting
internal analysis, e.g. names, titles, and combinations of titles and names:
(7) Randi White, our new headmaster, had also been busy.
(8) Christie's "Ten Little Niggers" is a detective story with no detective.
Categorization 351
(9) My dear Professor White, what can I do for you?

Note also the following phrases, which, superficially, seem to consist of a
noun plus a postmodifying adjective (typically of French origin): court
martial, heir apparent, Secretary General, devil incarnate, body politic, Poet
Laureate, president elect, etc. Unlike most other combinations of nouns and
adjectives, these collocations are syntactically fixed and receive main stress
on the adjective. It seems most appropriate to view them as compounds.
Disregarding the problem posed by compounds, heads in noun groups are
fairly easy to identify. Thus when noun groups function as subject, the
number of the head (singular or plural) governs subject-predicator concord:
(lOa) A cup of coffee is surely more expensive than a cup of tea.
(1 Ob) Two cups of coffee are surely more expensive than two cups of tea.
While this criterion is in general very reliable - even in cases where the head
is not necessarily the most important word - there are occasionally factors
(such as e.g. attraction, cf. section 7.6.5) which interfere with this neat
regular pattern. One phenomenon in particular requires mention here:
sometimes what looks like, and may well be analysed as, the syntactic head
does not in fact determine subject-predicator concord directly. Consider the
following examples (cf. Huddleston 1984: 236ff):
(11) A lot of milk was needed.
(12) A lot of eggs were needed.
In these examples concord seems to be governed by a part of the post-head
constituent (milk and eggs, respectively). If we stick to the criterion of
subject-predicator concord being governed by the head of the subject noun
group, we are forced to analyse a lot o/as a pre-head constituent (e.g. a
quantifier on a par with many and much), which is syntactically awkward. A
similar problem is posed by phrases like plenty of, lots of, the rest of, the
remainder of, a number of. Huddleston (1993: 87) suggests the term number-
transparent for these expressions: they let the number of the whole nominal
be determined by what is syntactically part of a post-head constituent, thus in
effect assuming number according to group-internal context.
Number-transparency sometimes affects determiner use in constructions
with kind of or sort of. e.g. these sort of theories, those sort of people, etc.
The noun NUMBER itself is number-transparent when it has a quantifying
meaning (typically with the indefinite article, as in e.g. a number of students
•were present). When it is used about a particular number as a number it
behaves like other nouns (typically with the definite article as in e.g. the
number of students enrolled has gone up a bit since we last talked).
352 Nominals
10.2.2. The semantics of nouns

Nominal heads, as well as single nouns, can be classified in many different
ways according to their meaning. In the introduction to this chapter we
defined the main communicative function of nominale as that of coding
meaning as things. But 'things' can be many, well, things! Nouns may be
interpreted in terms of a variety of both concrete things and abstractions:
physical objects (e.g. KNIFE, PIANO), persons (e.g. JOHN, PARENT),
animals (e.g. HORSE, DUCK), ideas (e.g. DEMOCRACY, ROMANTICISM),
relations (e.g. PARTNERSHIP, PARENTHOOD), emotions (e.g. LOVE,
SUSPICION), situations (e.g. ARRIVAL, RECITAL), etc. We can even talk
about things which do not exist in reality (e.g. ATLANTIS, UNICORN,
FAIRY). For a 'thing' to be encoded in language, all it takes is that we have
some mental representation of it. What is more, our understanding of a thing
may depend not so much on any inherent, objective features it may have as
on how we conceive of it. For example, 'a school* may be a 'physical object'
(as in They finished building the school last year), it may be a 'place' (as in
We walked all the way to the school), and it may be an 'institution' serving a
certain function (as in He went to school for 7 years only). What is more, our
subjective conception of 'school' in a particular context is sometimes
grammatically relevant, cf. the distinction between to the school and to
school in the examples above.
Another important distinction is that between human and non-human
referents of nominals. This distinction is relevant for the use of pronouns in
relation to nouns. Thus, when nouns with a human referent are postmodified
by a relative clause, the relative pronoun who is often selected (e.g. My
father, who retired last year, has moved to Brighton). By contrast, when
nouns with a non-human referent are postmodifed by a relative clause, the
relative pronoun which is often selected (e.g. The bike, which my son got for
his birthday, was stolen yesterday), cf. section 11.3.3. Also the choice of a
singular third person central pronoun (cf. section 11.1.2) relating to a
nominal depends on the human/non-human distinction: if the (head) noun
has a human referent, either the male term (he/him/his/himself) or the female
term (she/her/hers/herself) is used; if the (head) noun has non-human
reference, it is used:
(1) My neighbour didn't find his car key.
(2) The girl found herself m a bit of a mess.
(3) I looked for the dissertation, but I couldn't actually find it.
Interestingly there is sometimes vacillation according to the speaker's
disposition towards the referent: animals (especially domestic animals),
Categorization 353
ships, cars, countries, and other objects of human affection or concern may
be referred to (often endearingly) with the male or female term (e.g. Django,
my old labrador, was wagging his tail and She is an old Dutch schooner).
The following three sets of distinctions are examined in further detail
below: a) those pertaining to gender reflecting the sex of the referent of the
noun; b) those pertaining to what may be termed the referential scope of
nouns; and c) those pertaining to countability.
10.2.3. Gender
In many languages (e.g. Latin and German) each noun is marked specifically
for gender. The gender category typically comprises three members: the
masculine, the feminine and the neuter (in German, gender is expressed by
distinct determiners, e.g. der Mann, die Frau and das Buch (= 'the man1, 'the
woman' and 'the book', respectively). These three members sometimes bear
some relation to the sex of the referent (male, female or inanimate), but in
general this relation is very tenuous. In modern English there are no traces
left of the gender system of Old English. But the semantic distinctions
involved, especially between male and female, are occasionally expressed
morphologically. Below we review some of the ways to reflect sex distinc-
tions in nominals. Before we begin our presentation it is important to
remember that most nouns having human or animate referents are 'common
gender' or 'unisex': e.g. READER, NEIGHBOUR, DOCTOR, MUSICIAN, IN-
MATE, STUDENT, FRIEND, TEACHER, HELPER, EDITOR, FOOL, DRIVER,
PRISONER, EMPLOYEE, etc.
In some cases, English has one term for female referents, another for male
referents and a third for referents of either sex:
FATHER MOTHER PARENT
SON DAUGHTER CHILD
BOY GIRL CHILD
KING QUEEN MONARCH
BROTHER SISTER SIBLING
RAM EWE SHEEP
STALLION MARE HORSE
etc.
The male/female distinction is often expressed by unrelated words, as in the
trios above (cf. also UNCLE/AUNT, GENTLEMAN/LADY, MONK/NUN,
BACHELOR/SPINSTER, etc.). Occasionally, however, the distinction is
expressed morphologically (with -ess as the most common suffix):
354 Nominate
HERO HEROINE
ACTOR ACTRESS
LION LIONESS
MASTER MISTRESS
GOD GODDESS
etc.
In these examples, the male term is basic and the female term derived. There
are few exceptions to this dominant pattern:
WIDOWER WIDOW
BRIDEGROOM BRIDE
In some cases where we have related male and female terms, the male term
may be used as a unisex term, especially in contexts where the male/female
distinction is irrelevant:
LION LIONESS
TIGER TIGRESS
JEW JEWESS
In such cases, the male term is actually semantically unmarked while the
female term is semantically marked (positively female). This means that the
male term is only explicitly masculine when there is an overt contrast
involved (as in e.g. / saw both a lion and a lioness). Otherwise, the male
term has unisex reference (as in e.g. He shot three lions the other day). In the
following examples, the female is the unmarked term, having either feminine
or unisex reference, while the male term is explicitly masculine:
DRAKE DUCK
GANDER GOOSE
DRONE BEE
The male term MAN is special in denoting either 'mankind1 in general (thus
including women, as in e.g. All men are equal) or simply 'male members of
the human race' (as in Men are generally taller than women and These men
are exceptionally tall).
When MAN forms a part of a compound, it sometimes denotes male (as in
MANSERVANT, BUSINESSMAN, DOORMAN), sometimes it has unisex
reference (as in MANSLAUGHTER, SPOKESMAN, STATESMAN, CHAIR-
MAN). But there is a tendency to avoid using MAN compounds about
women: CHAIRPERSON, STATESWOMAN, SPOKESWOMAN.
As Jespersen (1933: 192) points out: when assuming complement function,
male terms are sometimes used more readily about women:
(1) She was a master of the situation.
Categorization 355
(2) Ann was always a lover of beautiful art.

In addition to man and woman, male anafemale are often used to specify sex
in neutral nouns: male/female reader, male/female driver, etc. Note also:
gentleman friend, lady friend, boyfriend, girlfriend, maid-servant.
Note finally that special compounds are sometimes used of animals to
specify sex: COCK-PHEASANT / HEN-PHEASANT, JACKASS / JENNY-ASS,
BILLY-GOAT / NANNY-GOAT, TOM-CAT / TABBY-CAT.
10.2.4. Types of nouns and referents

Let us begin this section by offering a note on our use of the terms 'referent'
and 'reference' in connection with nominals. Both terms are used in a fairly
broad sense: 'referent' about our mental representation of a thing coded in an
expression and 'reference' about the communicative function of establishing
something as a referent. Thus, for example, when a speaker says / bought a
new car yesterday, our conception of the car that he bought is the referent
and a new car is a referring expression, and the link between them is one of
reference. In other words, whenever we use 'referent' we emphasize the
nature of the thing expressed and whenever we use 'reference' we emphasize
the way we use language to code things.
Central to the notion of categorization are the following main kinds of
referents: unique, generic and class-member referents. Though, from a
strictly objective point of view, every thing in the world is unique in some
sense, as human beings we tend to classify things sharing one or more
similarities as belonging to the same type or class. Thus apart from persons
or entities that we recognize as unique (such as Peter Schmeichel, Paris,
Spain, etc.), we have cars, bikes, books, trees, etc. Every car is, strictly
speaking, unique and yet the noun CAR can be used to refer to a fascinating
range of vehicles. Any particular car is thus a member of the class of things
which may be appropriately referred to by the noun CAR. Most of the
examples offered so far in this chapter are examples of class-member
referents in this sense: the (head) noun is used to refer to one or more
particular things that may be conceived of as members of a class of things
which may be appropriately referred to by using the noun. For example, in an
expression like I found the dissertation in the top drawer, the referent of the
direct object the dissertation may be understood as 'a (specific) member of
the class of things appropriately termed dissertation'', and in an expression
like / want an apple, please, the referent of an apple may be understood as 'a
(non-specific) member of the class of things appropriately termed apple''.
Whether specific or non-specific, the referent is viewed as an instance of a
more general kind or type ofthing (dissertations, apples).
356 Nominate
Significantly, instead of referring to individual class members, we may

choose to refer to the kind or type in question as such: e.g. The funnel-web
spider is very common in New South Wales and Young children need a lot of
attention. Here The funnel-web spider refers to a kind of spider (in contrast
to an example like Jack killed the funnel-web spider with his spade, which
refers to a particular member of the class of funnel-web spiders), and Young
children refers to the whole class of young children (in contrast to The young
children missed their parents, which refers to a group of particular young
children). In both cases we have a so-called generic referent.
Corresponding to the semantic distinction between unique referents and
non-unique (generic or class-member) referents there is the form distinction
between proper nouns and common nouns. Proper nouns are capitalized, and
with their central function - that of naming - they are fairly restricted with
respect to determination and quantification: Mary, Sweden, London, the
Hebrides, *a Mary, *the Sweden, *some Londons, *these Hebrides. Com-
mon nouns, by contrast, are fairly unrestricted with respect to determination
and quantification: a/the/these/some trains, a/the/these/some pencils, etc. The
distinction between generic and class-member referents is also grammatically
relevant. It is not, however, reflected in different classes of nouns but, as we
shall see, in the use of determiners.
There is no strict one-to-one correspondence between unique referents and
proper nouns, on the one hand, and between non-unique referents and
common nouns on the other. Following grammarians like Huddleston (1984
and 1988), we distinguish between proper nouns and names. Proper nouns
are typically, though far from inevitably, used as names (e.g. Jack, Germany,
etc.). But names may consist, wholly or partially, of common nouns: e.g.
High Street, Congress, the Copenhagen Business School, London Bridge,
Mother, Uncle, etc. And when not used simply for naming, proper nouns are
far more unrestricted with respect to determination and quantification:
(1) He can be a real Sylvester Stallone sometimes.
(2) It was good fun to watch all the young Peter Schmeichels practising.
In these examples, the proper noun expressions all the young Peter
Schmeichels and a real Sylvester Stallone have class-member referents rather
than unique referents, assigning to their referents certain ambitions, qualities,
status or behaviour that we associate with the bearer of the name.
10.2.5. Countability
The referents of common nouns are subject to a further distinction between
countable and non-countable. Nouns whose referents are conceived of as
Categorization 357
something individualized we can count are called count nouns whereas

nouns whose referents are conceived of as something »»individualized we
cannot count (or simply do not count) are called mass nouns. Here are some
examples of count nouns: BOOK, WINDOW, CAR, PENCIL, HOUSE, BOAT,
IDEA, FRIEND, etc. And here are some examples of mass nouns: WATER,
SAND, BUTTER, MILK, RICE, MONEY, FURNITURE, ADVICE, NEWS, PER-
MISSION, etc. The possible referents of most of these are obviously non-
countable but, strictly speaking, e.g. rice and money are countable. In
English, however, the nouns RICE and M O N E Y are mass nouns by
convention.
Count nouns allow of quantification in terms of the singular/plural
distinction (e.g. book/books, window/windows, etc.) and are thus compatible
with pronominal determiners and quantifiers like some, more, many, few,
several, etc. and incompatible with much and less: many ideas, few move-
ments, *much computers, *less schools (the not infrequent use of less in
connection with count nouns, as in less schools, less problems, etc., is
generally considered colloquial). Mass nouns, by contrast, allow of direct
quantification only in terms of pronominal determiners and quantifiers: the
singular/plural distinction does not apply to mass nouns. They are compatible
with much, less, little, some and more but not with many, several, few and
one or with the indefinite article a(n): much water, less wine, little sand, *few
furniture / * a furniture / *'manyfurnitures.
However, mass nouns, whether concrete or abstract, may be quantified
more indirectly in partitive »/-constructions preceded by a quantified count-
noun: two pints of bitter, a cup of coffee, many bottles of wine,few bowls of
rice, several slices of bread, an acre of land, a word of advice, a fit of
passion, an attack of pneumonia, etc. In some such partitive expressions
there is a very close relation between the head count noun and the quantified
mass noun: a suit of armour, a tankard of beer, a sheet of paper, a clove of
garlic', a stroke of luck, a pang of remorse, a flash of lightning. Note that
count nouns may be quantified in a similar fashion: a page of a book, etc.
Very similar to quantity partition is quality partition as in a new kind of
butter and a new generation of computers.
It is important to note that the distinction between count nouns and mass
nouns is somewhat blurred, grammatically speaking. Many nouns are used
equally well as count nouns (for bounded entities) and mass nouns (for
unbounded material or concept), e.g. STONE, COCONUT and CONVICTION:
(la) He found three stones.
(1 b) The figure was of s tone.
(2a) There were coconuts everywhere on the beach.
358 Nominate
(2b) The worshippers bought coconut and flowers for their offerings.
(3a) He acted in accordance with his convictions.
(3b) It appeared to be the result of blind conviction.
Many nouns that primarily behave like count nouns are sometimes used the
way mass nouns are used, and vice versa:
(4) Everything was grimy under a low ceiling of grey cloud.
(5) One black sock had sagged to reveal a section of bare leg.
(6) There was enough moon now to silver the minarets outside.
(7) This is actually an excellent wine.
(8) He imports several coffees from Africa.
(9) Over the years she did me many kindnesses.
(10) Two coffees, please.
When typical count nouns, such as CLOUD, LEG and MOON, are used like
mass nouns (as in (4) to (6)), the speaker or writer emphasizes the material,
character or concept of the referent rather than simply the referent as a
bounded entity (thus e.g. moon in (6) gets very close in meaning to
MOONLIGHT). Conversely, when typical mass nouns are used like count
nouns, as in (7) to (10), the expression either has sub-generic reference, i.e. it
denotes a subclass of a class, as in (7) and (8) (an excellent wine = an
excellent kind of wine, several coffees = several kinds of coffee), or it denotes
instances or realizations of the non-countable entity (many kindnesses = e.g.
many acts of kindness, two coffees = two cups of coffee) as in (9) and (10).
The concept of countability does not usually apply to proper nouns
functioning as names: they are neither count nouns nor mass nouns. Though
names are either formally singular (e.g. John Wilson, London, France) or,
less often, formally plural (e.g. the Hebrides, General Motors), we do not
immediately conceive of their referents as countable, or even quantifiable.
However, like count nouns, names typically have individualized referents.
With a more extensive function than simply naming, proper nouns readily
accept quantification: all the young Peter Schmeichels, a real Sylvester
Stallone, etc. But there are also examples where proper nouns retain their
status as names despite the association of countability:
(11) Have you invited the Wilsons to stay with us?
(12) Is there a Sarah Mortimer staying at this hotel?
Here the Wilsons means 'the Wilson family' and a Sarah Mortimer means 'a
certain Sarah Mortimer or 'someone called Sarah Mortimer'. In both cases
there is a clear sense of uniqueness despite the explicit quantification.
Consider finally examples like:
Categorization 359
(13) How many Peters are there in this department?

(14) Where does he spend all his Christmasesl
(15) We play badminton on Mondays.
Here the proper noun expressions Peters, Christmases and Mondays have
referents actually called 'Peter', 'Christmas' and 'Monday', respectively (and
thus differ from the proper noun expressions Peter Schmeichels and a real
Sylvester Stallone). Such examples combine classmembership and unique-
ness by referring to 'unique but recurring phenomena'. What these examples
also show is that there is a grey zone between common nouns and proper
nouns, and that this affects the question of countability.
10.2.6. Recapitulation
Let us briefly recapitulate the findings of the last few sections on cat-
egorization. As we have seen, there is no clear-cut definition of what exactly
constitutes the head in a noun group. A practical working definition is that it
is a single noun or syntactically fixed compound with unitary stress and a
noun as the key element. The function of the head is to categorize the
referent. This categorization can be described in terms of a number of
general semantic distinctions (e.g. human/non-human, male/female/unisex/
inanimate), the referential scope of nouns (unique, generic and specific/non-
specific class-member referents) and countability (countable vs. non-
countable). These distinctions are grammatically relevant in a number of
ways: e.g. morphological derivation, choice of co-occurring pronouns,
determiner usage and classification of nouns into subtypes (proper vs.
common nouns; count vs. mass nouns, singular vs. plural nouns).
The following charts summarize the classification of referents and nouns:
(1)
, unique
Referents /
non-unique
(2)
χ proper
Nouns <^
common
360 Nominate
10.3. Determination
10.3.1. Types of determiner
Determiners are used to signal the kind of reference involved in the
expression of a nominal. There are four main subcategories of determination:
(i) Definite determination. There are four form types realizing definite
determination: a) the definite article the (as in the doctor, the bright girls); b)
the demonstrative pronouns this, that, these, those (as in this bicycle, those
bastards); c) the possessive pronouns my, your, his, her, its, our and their (as
in my wedding, his student days); and d) genitive nominals (as in Jack's
truck, my old father's idea). To these types of definite determination we may
add such, which is a demonstrative-like qualitative pronoun with subgeneric
meaning: such misery, such students.
(ii) Indefinite determination. There are three form types expressing
indefinite determination: a) the indefinite article a(n) (as in a_ new hall, an
arrogant journalist); b) zero (0) (as in _ professors, _ sugar); c) the
indefinite pronouns any, no, each, every, either, neither, some (as in any
suggestion, UQ_joy, either way, some girl(s)). To these indefinite determiners
we may add one as an emphatic alternative to the indefinite article and
another, which combines the indefinite article an and the modifier other (cf.
the definite counterpart, which is in two words: the other).
(iii) Interrogative determination. Interrogative determination, which is
often indefinite in character, is used to form a question about the head. There
are three interrogative pronouns which may serve a determinative function:
which, what and whose (as in Which book do you prefer? / What solution did
she come up with? /Whose key is this?).
(iv) Relative determination. Which, what and whose may also serve as
relative determiners with more definite meaning (as in Her visitor left at four
o'clock, by which time the FBI had already arrived /He enjoyed what wine
was left / The boy whose bike was stolen knocked on my door). In addition
we have emphatic whichever and whatever (as in Whichever book you
choose I am sure your parents will approve / Whatever solution he comes up
with she will support him). Relative determination by which and whose is
used to relate the head to a preceding constituent. While whose is stylistically
unmarked, which is rather formal as a determiner. What(ever) and whichever
are in this function independent relative pronouns (cf. section 5.4.1).
We do not regard the first- and second-person plural personal pronouns in
examples like the following as determiners:
Determination 361
1l) What are you guys doing on Sunday?

(2) We Americans believe in democracy.
(3) They simply hate us Germans.
In such examples, the noun specifies the meaning of the pronoun and is thus
best analysed as an elaborative dependent in a pronoun group.
Note that the noun could be replaced with a different form type without
changing the overall construction type; compare e.g. (2) with (4):
(4) We in the English Department believe in tough grammar exams.
This kind of construction should not be confused with the non-standard,
colloquial use of them as an alternative to those in examples like:
(5) We don't like them bastards.
We also exclude from our class of determiners indefinite pronouns like all,
both and one. These pronouns, which are quantifiers rather than determiners,
will be briefly dealt with in section 10.3.3.
Here is an overview of the determiners identified in this section:
definite article: the
demonstratives: this, that, these,

definite those, (such)
possessive pronouns: my, your, his,

her, its, our, their
genitive constructions: John's, women's,

the old man's, etc.
Determination indefinite article: α (η)
indefinite zero: 0
indefinite pronouns: any, no, each,

(n)either, every, some, (one, another)
interrogative — which, what, whose
relative whichever), what(ever), whose

362 Nominal*
10.3.2. Co-occurring determiners: pre- and postdeterminers

Determiners are normally in a paradigmatic choice relation (cf. section 2.8)
and thus in complementary distribution: if we select one, we cannot also
select another. There are, however, certain exceptions to this restriction:
(i) such may precede the indefinite article plus singular count noun (as in
such afooL such a good idea);
(ii) what, too, may precede the indefinite article plus singular count noun to
form an exclamative expression (as in fflhat a fool! / What a good idea);
(iii) possessive pronouns and genitive constructions may be followed by
every plus a singular count noun as an emphatic alternative to an expression
with the quantifier all plus a plural count noun (compare the old man's every
move /all the old man's moves, her every wish /all her wishes);
(iv) such occasionally follows a determinative indefinite pronoun (as in no
such luck, any such move, each such development, some such problem):
(v) in elevated, rhetorical speech, possessive pronouns sometimes follow
demonstrative pronouns (as in e.g. On this his last day in office he visited his
predecessor). One possible analysis of this kind of construction, however, is
to treat the possessive pronoun as a nonrestrictive specifier of the demon-
strative pronoun rather than a separate determiner (On this his last day ... =
On this last day...).
The following chart summarizes the possible collocations of determiners:
DETERMINERS:
pre- central post- EXAMPLES
such such a fool

r indefinite article
what j what a fool
genitive constructions Jack 's every move

f every
possessive pronouns J her every wish
indefinite pronouns such any such luck
(demonstrative pronouns) (possessive pronouns) (this his last day)

Determination 363
What and such are predeterminers when they precede the indefinite article
and every, such and possessive pronouns are postdeterminers when they
follow other determiners. All other determiners are central determiners.
When what, such and every are the only determiners in noun groups (as in
such misery, every woman, etc.) they, too, are called central determiners.
10.3.3. Determiners and quantifiers

Determiners are inextricably linked up with quantifiers. Thus, obviously,
demonstrative pronouns are either singular (this, that) or plural (these, those)
in concord with the head noun, and the indefinite article (as well as its
emphatic alternative one) is used only with singular count nouns (in contrast
to 0, which is used with plural count nouns and with mass nouns). Also the
indefinite determiners any, no, each, every, either, neither, some have a more
or less clear association of quantification about them: in some of their central
uses they add different nuances of quantifying meaning to the basic
indefinite class-member or mass referents of the noun group as a whole. Of
the indefinite determiners, no and some are the ones with the clearest
association of quantification: no means 'absence of quantity1 and some means
'indefinite, undefined amount or number of. At the same time, however, no
and some are clear determiners. When followed by e.g. a singular count
noun, no is the negative form of the indefinite article (= 'not α(η)*); compare:
(1) There is a pen in the top drawer.
(2) There is no pen in the top drawer.
When followed by e.g. a plural count noun, some is the plural equivalent of
the indefinite article in noun groups with class-member reference; compare:
(3) I met a professor in London.
(4) I met some professors in London.
Some thus supplements 0, which in connection with plural count nouns often
signals generic reference; compare:
(5) Professors like poetry.
(6) Some professors like poetry.
While any, no, each, every, either, neither and some are primarily deter-
miners with an association of quantification, other indefinite pronouns and
numerals are primarily quantifiers with an association of determination: e.g.
both, all, half, one, many, five, second, etc. Note that unlike the indefinite
pronouns listed as determiners above, all these items are compatible with
both definite and indefinite reference: both girls/both the girls', all cars /all
the cars; half a bottle / half the bottle; one case / the one case; many books /
364 Nominals
the many books; five calls / the five calls; the second attempt/ a second
attempt. While many of these items may occur in constructions without
genuine determiners, they can all co-occur with central determiners. Like
predeterminers, the following quantifiers may precede central determiners:
(i) Both, all, half, these three quantifiers may precede definite central
determiners (the definite article, demonstrative pronouns and genitive/
possessive constructions). They can also function as heads in pronoun groups
with a very similar meaning; compare: all the soldiers /all of the soldiers;
half the money / half of the money; both these solutions / both of these
solutions. Unlike the two others, half may precede the indefinite article in
connection with head nouns expressing quantity or measurement: half α
pound, half a mile, half a pint, half an inch.
(ii) Multipliers: double, twice, three times, etc. may precede definite central
determiners (just like both, all and half): double the average, twice his
income, three times this amount. In expressions of frequency where the head
noun expresses a standard against which the frequency is determined, once,
twice and expressions with times (e.g. three times) may precede the indefinite
article or the indefinite pronouns every or each', once a week, twice each
month, three times every fortnight.
(iii) Fractions: two-thirds, one-fifth, etc. may precede the definite article
(e.g. two-thirds the amount) or serve as group heads followed by an of-
construction (e.g. two-thirds of his salary).
Like postdeterminers and modifiers, many quantifiers may follow central
determiners:
(i) Cardinal numbers: one, two, three, four, etc. may follow definite central
determiners (his one objection, these two claims, the three pencils, etc.).
Cardinal numbers may also serve as group heads followed by an of-
construction (one of his objections, two of these claims, three of the pencils).
(ii) Ordinal numbers and other ordinals: first, second, third, fourth, etc.;
next, last, other, further. These items may follow definite central determiners
(the first attempt, his second car, this third meeting, etc.). When ordinal
numbers serve as classifiers, they may follow indefinite central determiners
such as the indefinite article and each: a first chapter, a second attempt; each
first chapter, each second attempt etc. In expressions of frequency, ordinal
numbers follow every: every second meeting, every third visitor, etc.
(iii) Other quantifiers: many, (a) few, several, various, more, most, (a)
little, less and others may follow central definite determiners: the many
problems, his few friends, John 's several attempts, these various solutions,
Determination 365
etc. They also serve as group heads followed by an «^construction: many of

the problems, few of his friends, etc. Only many may precede the indefinite
article in cases like many a kiss (which is close in meaning to many kisses
but has a singularizing effect).
Of the determiners identified in section 10.3.1 above, we shall now examine
more closely the definite article (in section 10.3.5), the indefinite article (in
section 10.3.6), zero (in section 10.3.7) and the genitive construction (in
section 10.3.8). Pronominal determiners (demonstratives, indefinites,
possessives, interrogatives) will be dealt with more thoroughly in chapter 11.
10.3.4. Referential orientation

There are two ways in which it is relevant to speak of types of reference (as
distinct from types of referents, cf. section 10.2.4): a) types relating to the
question of 'referential orientation' of referring expressions (i.e. the question
of where to look for a referent), and b) types relating to the relationship
between referring expression and type of referent (e.g. 'definite specific
reference1 = reference to a specific referent by means of a definite nominal).
In this section we examine the former typology, leaving the latter to our
discussion of the use of the articles in sections 10.3.5-7.
Let us begin by distinguishing two main referential orientations. There is
reference to something mentioned elsewhere in the (spoken or written) text:
this is called textual (or endophoric) reference (note that 'text' and 'textual'
are to be understood to include not only written language but also spoken
language). And there is reference to something in the extralinguistic, non-
textual context, i.e. something in some real or fictional world: we call that
non-textual (or exophoric) reference. Here are first some possible examples
of non-textual reference:
(1) Will you pass the salt, please.
(2) There's an apple in the basket.
(3) Hurry up, or you will run into their security guard.
(4) The sun set about half an hour later.
These examples are perfectly well-formed even if there is no other mention
of the referents in the preceding or following text. As we see, the context
may be very specific, as in (1) to (3), where the referent is close by and can
be manipulated in one way or another, or it may be very general, as in (4),
where the referent is a natural phenomenon on which human existence
depends, and thus actually part of any context if only very implicitly.
There are two main types of textual reference: anaphoric and cataphoric
reference. Anaphoric reference is backward reference to a preceding textual
366 Nominals
unit (called the antecedent), whereas cataphoric reference is forward

reference to a following unit (there is no appropriate traditional name for this
unit, but let us call it 'postcedent'). Here are some examples of anaphoric
reference (in which the antecedent/postcedent is indicated in curly brackets):
(5) {A man} and a woman entered. The man was wearing a tie.
(6) If the movement is to preserve its appeal, {radical feminism} must realize
that the ideological climate has changed.
Textual reference, both anaphoric and cataphoric, may cut across the
sentence boundary: in example (6) the relation is intrasentential, in example
(5) it is extrasentential.
Anaphoric reference is repetitive if the head noun of the referring group is
identical with the head noun of the antecedent, otherwise it is non-repetitive.
Thus while (5) above is an example of repetitive anaphoric reference, the
following sentences are examples of non-repetitive anaphoric reference:
(7) {A man and a woman} were sitting on the bench; the couple seemed to be
very much in love.
(8) For several weeks I avoided {Roger}. I couldn't bear to see his gleeful face.
God, how I hated the bastard.
Cataphoric reference is always non-repetitive (cf. example (6)).
Non-repetitive anaphoric reference is either direct or indirect (repetitive
reference is always direct). The examples looked at so far are instances of
direct reference in the sense that the expression under analysis establishes
(repetitively or non-repetitively) exactly the same entity as the antecedent.
With indirect reference, the expression establishes a referent which is related
to, but not identical with, the antecedent. The antecedent provides a
background against which the existence of the referent of the expression
under analysis may be recognized or accepted, as in:
(9) It was {a very wide ditch}, and when they crept up to the edge and looked
into it they could see it was also very deep, and there were many rocks at the
bottom, (for discussion of this example, see Kvistgaard Jakobsen 1994: 65f)
The italicized noun groups are examples of indirect anaphoric reference with
a very wide ditch as the antecedent: the edge and the bottom are to be
understood precisely as 'the edge of the ditch' and 'the bottom of the ditch',
respectively.
The following chart summarizes the types of reference identified above in
connection with referential orientation:
Determination 367
. , ι
textual
/
(
(or endophoric) \
,
, anaphoric (
f repetetive
non-repetitive \
\
. .
/ direct
indirect
U
.
intra-
. sentential
extra-
sentential
cataphoric — non-repetitive
Reference
specific
non-textual
(or exophoric)
general
10.3.5. The definite article

Historically, the definite article the is a weakened demonstrative form. In
many of its uses, this is semantically evident: the definite article typically
singles out or delimits the referent of the noun group relative to the commun-
icative context. Syntactically, the definite article is a central determiner and
thus precedes any modifiers in the noun group:
(1) He is the right man for the job.
(2) He gave the very same talk last week.
However, certain degree adverbs and pronouns like QUITE and MUCH may
intensify the noun group as a whole, especially in expressions involving
comparison, in which case they precede the definite article:
(3) He is not quite the man he used to be.
(4) He gave much the same talk last week.
When nouns are coordinated, article usage varies:
(5) The Danes and Swedes felt humiliated.
(6) The Danes and the Swedes felt humiliated.
In (5) it is possible to interpret the two nouns as more closely connected (i.e.
as referring to a group of Danes and Swedes). In (6) the repetition of the
definite article secures a separate focus on the two nouns (i.e. there are two
separate groups involved: a group of Danes and a group of Swedes).
When modifiers are coordinated, each receives a separate focus if the
definite article is repeated. In noun groups with a plural noun, this repetition
of the definite article sometimes results in distributive meaning with the
modifiers relating to different referents of the noun; compare:
368 Nominals
(7) He was by far the kindest and the most competent officer, (non-distributive)
(8) He handed me the yellow and the red boxes, (distributive)
In the last example, we can only suppress the association of distribution by
leaving the second article out: He handed me the yellow and red boxes (i.e.
the two-coloured boxes). However, not repeating the definite article is not
always enough to exclude a distributive interpretation:
(9) No, I'm talking about the eighteenth and nineteenth centuries.
In the rest of this section (which is in part based on Schibsbye's treatment of
the topic and his examples (1970: 219ff)), we shall review the various more
specific uses of the definite article:
A) Definite specific reference. In its most central use, the definite article is
used by the speaker in connection with common nouns to indicate that he
expects the listener to be able to single out a particular referent, typically on
the basis of a shared familiarity. For example, if someone says: Have you
seen the kettle?, the speaker signals that he expects the kettle in question to
be familiar to the listener. If the speaker thinks that the listener may not
know about the kettle, he will use the indefinite article: Have you seen a
kettle around here somewhere? If the speaker thinks that the listener has
some knowledge of the possible referents of the noun KETTLE but not
enough to recognize the particular kettle in question, he will supply what
information he deems necessary for the listener to do so. In other words, he
will establish a shared familiarity, e.g.:
(10) Have you seen the kettle that I borrowed from Sally!
(11) Have you seen the kettle with the broken handle?
(12) Have you seen the new kettle?
In example (10), the shared familiarity is provided by the restrictive relative
clause. Put differently, the restrictive relative clause warrants the use of the
definite article by expressing information which will enable the listener to
recognize what the speaker is talking about. In example (11), the definite
article is warranted in this sense by the postmodifying preposition group with
the broken handle, and in example (12) it is warranted by the premodifying
adjective new. In other words, by using the definite article the speaker
signals that he will provide the listener with enough information to figure out
what the speaker is talking about. Sometimes the speaker is wrong, in which
case a dialogue like the following may take place:
(13) Speaker A: Have you seen the kettle?
Speaker B: What kettle?
Speaker A: The one I borrowed from Sally.
Determination 369
Note here the interrogative determiner What in speaker B's question.

The degree of precision with which the listener is expected by the speaker
to recognize the referent may vary contextually. In some cases, the definite
article merely signals a plea for the listener's acceptance of the existence of a
particular referent. For example, in a case like It turned out that John had
been to the same school as Max (cf. Huddleston 1984: 249), the speaker
expects no more of the listener than his or her ready acceptance of the
existence of a particular school to which both John and Max used to go.
The kind of reference involved when the definite article is used by the
speaker to indicate that he expects the listener to be able to single out a
particular referent is called definite specific reference. Definite specific
reference is possible with all four main kinds of nouns (common count
singular, common count plural, common mass, proper):
(14) Dan argued that Owen truly adored the school.
(15) The girls next door were doing their homework.
(16) She passed him the sugar.
(17) I want the two Peters in this class to report to Mr Wilson.
Characteristically, definite specific reference picks the referent out as one or
more particular members of a class (as in examples (14), (15) and (17)) or as
particular, limited 'sub-mass* (as in example (16)). In each case, however,
there is a clear implication that, in context, the referent is unique (if the noun
is singular) or all-inclusive (if the noun is plural). Thus in the context of
example (14) there is only one school to which the expression the school
may apply and in the context of example (15), all and only the girls next door
are included in the referent of The girls next door.
Reference to specific limitations or specific bounded instances of non-
countable entities such as e.g. 'love', 'life', 'nature', 'goodness', 'tyranny', 'art',
'materialism', 'mud', 'water', 'rice' etc. requires the definite article; compare:
(18a) He studied architecture.
(18b) He studied the architecture of the Roman Empire.
(19a) Life is sweet.
(19b) The new life she gave me was so exciting.
Non-specific limitation of a generic referent is often possible without the
definite article, as in:
(18c) He studied Roman architecture.
The result of such limitation is sub-generic reference: in (18c) the referent
"Roman architecture' is a subtype of'architecture'.
370 Nominate
While definite specific reference is the central function of the definite article,
there are other important uses:
B) Definite 'non-specific* reference. Sometimes the definite article is used
in cases where there is no specific referent yet but simply an expectation or
assumption that there will be a specific thing answering the description of the
definite expression as a whole. Such cases of potential but unrealized
specificness is sometimes called definite non-specific reference (despite the
fact that the non-specific meaning is a result of contextualization rather than
linguistic expression):
(20) He is still looking for the right girl to marry.
(21) I want you to reach the cheapest possible solution.
(22) The winner of the final tomorrow morning will receive £2,000.
As these examples indicate, definite non-specific reference is especially
common in constructions involving (implicit or explicit) comparison and/or
future situational reference.
C) Generic reference. In the singular, generic expressions may take the
definite article:
(23) He took a professional interest in the diesel engine.
(24) The funnel-web spider is common in New South Wales.
To use a singular count noun representatively for the whole class or kind in
this way is often somewhat more formal than to use indefinite plural
expressions (e.g. diesel engines, funnel-web spiders). With musical instru-
ments and dances, however, the definite singular noun group is the usual
expression:
(25) She plays the guitar and the lute.
(26) Jack absolutely hates to dance the foxtrot.
In the plural, generic reference is typically indefinite: diesel engines ('all
engines'). Normally the definite article is used only when there is reference to
national or ethnic groups (e.g. the Russians, the Europeans, the Blacks, etc.).
For generic expressions with the definite article in adjective groups (like
the rich, the young, etc.) cf. section 12.4.2.
D) Unique reference. Since names have unique referents, they usually lack
the definite article (Jack, London, France, Europe, Carlsberg, etc.). There
are, however, a number of exceptions to this rule:
(i) Some geographical names, especially plural names of regions,
archipelagos and mountain ranges, take the definite article: e.g. the Hague,
the Tyrol, the Sahara, the Ukraine; the Americas, the Orkneys, the Alps. The
Determination 3 71
same applies to names of seas, rivers and canals: the Pacific, the Atlantic, the
Thames, the Nile, the Suez Canal, etc. The article is dropped in a river name
if it is part of the name of a town (as in Newcastle-on-Tyne) or if it is part of
an enumeration (as in a network of canals, connecting Humber, Severn,
Mersey and Thames, cf. Schibsbye 1970: 22If).
(ii) Names of hotels, restaurants, clubs, cinemas, theatres, major buildings,
journals and ships often require the definite article: the Imperial, the Hungry
Monk, the Savoy, the Taj Mahal, The Times, the Estonia, the Mayflower, etc.
(Hi) Some titles of persons: the Queen of Denmark, the Reverend Roger
Smith, the President of the United States, the Marquess of Salisbury, etc.
(iv) Proper nouns used as names may take the definite article if they are
(restrictively or non-restrictively) modified to express a certain aspect or
version of the referent: the young Churchill (= "Churchill as a young man1 /
'Churchill Junior' or 'Churchill, who was a young man', depending on young
being restrictive or non-restrictive), the real Spain, the famous Mick Jagger,
etc. Expression like the young Churchill and the famous Mick Jagger may
also be interpreted to refer restrictively to particular bearers of these names in
contrast to other bearers of the same name, e.g. someone called Churchill
who is not young and someone called Mick Jagger who is not famous. The
article may be dropped if the group assumes the status of a new name (as in
Merry Old England and Ancient Ireland) or if the speaker or writer wishes to
express what Schibsbye (1970: 227f.) calls "benevolent interest", as in:
(27) The fall is that of famed Niagara, the roar awe-inspiring.
(28) When Jess Conrad kissed 11-year-old Susan Pinkney, pretty Yvonne Kersting
broke down.
Such examples are particularly common in journalism.
(v) Proper nouns used for specific, limited (typically restrictively modified)
class-member reference:
(29) Mr Grossman is the Burke of our day.
(30) Stockholm is the Venice of the North.
(31) I would like to speak to the two Peters in this class.
In the first two examples, the names Burke and Venice are used qualitatively
about referents with other names. In the last example, Peters is used
quantitatively about specific members of the class of people called 'Peter'.
(vi) Note in particular the use of the definite article in connection with the
names of weekdays, months, annual events, etc. Without the definite article,
such nouns are used deictically, i.e. in relation to the 'here and now' of the
372 Nominals
speaker, whereas with the definite article, they are used in relation to some
other point of time relevant in the context; compare:
(32) I will see you on (next) Monday,
(33) I will see you on the (following) Monday.
(34) She fell in love with him last August.
(35) She fell in love with him the following August.
Without the article, such nouns have unique referents, with the article they
have definite specific class-member referents.
It is interesting to note that in many names the presence of the definite
article is caused by the name being partially or fully derived from common
nouns or constructions containing common nouns as the head of the group:
the Pacific (Ocean), the (River) Thames, the Gulf (of Mexico), the Täte
(Gallery), the Sahara (Desert), etc.
E) The emphatic definite article. Note finally the emphatic use of the
definite article (pronounced /oi:/) to denote that the referent deserves the
description provided by the group in the highest degree:
(36) He is the expert on computational linguistics.
(37) She is the master of modern dance.
10.3.6. The indefinite article

We turn now to the indefinite article. As pointed out earlier, the indefinite
article a(n) is used only in connection with singular nouns. Historically, it is
a reduced form of one and in some examples it is close to having the
meaning of a numeral (cf. Jespersen 1933: 174f.):
(1) In a word, I don't like him.
(2) Rome wasn't built in a day.
(3) I would like two cheeseburgers and a coke, please.
The indefinite article is a central determiner and thus usually precedes any
modifiers in the noun group:
(4) Farrokh was an unassimilated Canadian.
There is, however, attraction of modifiers to pre-determiner position by
HOW(EVER), AS and SO in examples like the following (cf. the use of such
and what as predeterminers mentioned in section 10.3.2):
(5) However beautiful a woman she is, she does not fool me.
(6) He may not be as competent a doctor as Bill.
(7) I cannot resist so nice a proposition.
Determination 373
Attraction is also possible (with little or no difference of meaning) when

modifiers appear in conjunction with the degree adverbs TOO and NO LESS:
(8) Sandra is a too critical reader.
(9) Sandra is too critical a reader.
Adverbs like QUITE and RATHER are often found in pre-determiner position
relating to all of what follows in the noun group rather than simply as a
dependent of a modifier; compare:
(10) Bill is α quite/rather competent sailor.
(11) Bill is quite/rather a competent sailor.
In connection with coordinated nouns, the indefinite article is usually
repeated if the two nouns refer to separate entities, qualities, aspects, etc. If
there is felt to be a close relation between the two nouns, the indefinite article
is used only before the first noun; compare:
(12) I asked for a knife and fork.
(13) Dr. Daruwalla was an orthopedist and a Duckworthian.
In connection with coordinated modifiers, the indefinite article is not
repeated except to create emphasis on different aspects of the referent or if
the modifiers relate distributively to different referents; compare:
(14) She was a stunningly beautiful and an intelligent woman, (non-distributive)
(15) This was an acceptable, even an honourable, tradition, (non-distributive)
(16) I gave him a red and yellow box. (non-distributive)
(17) I gave him a red and a yellow box. (distributive)
Like the definite article, the indefinite article is used for a number of deter-
minative functions:
A) Unmarked determination: class-member reference. Basically, the
indefinite article is used by the speaker in connection with singular count
nouns with class-member referents to indicate that he or she does not expect
the listener to be familiar with (and hence to be able to single out) a
particular referent of the noun group. In this way, the indefinite article is
used when the conditions for using the definite article are absent:
(18) She was stuck between a grim-looking American and a grim-looking
Scandinavian.
(19) He was seated in the darkness in a gazebo by the lakeshore, quietly listening
to a recording of Kierkegaard's Either/Or.
(20) She was wearing an enormous full-length kimono-style dress.
(21) You were stepped on by an elephant while your father was buying
cigarettes?
374 Nominate
In each case, the noun group containing the indefinite article refers to a
member of the class of things potentially referred to by the head noun: for
example, an elephant refers to a member of the class of elephants.
Once the speaker or writer has introduced a referent by an indefinite noun
group, enough familiarity has been established to warrant the use of the
definite article in subsequent noun groups with the same referent (anaphoric
repetitive or non-repetitive reference):
(22) "You were stepped on by an elephant while your father was buying
cigarettes?', Farrokh asked ... The doctor didn't believe he could fix what the
elephant had done.
(23) A badly limping boy could occasionally be seen standing on his head at
Chowpatty Beach. The doctor knew that this wasn't a trick of sufficient
promise for Vinod and Deepa to offer the urchin a home at the circus. The
boy had slept on the beach ...
B) Description in terms of class-membership. Noun groups containing the
indefinite article are particularly frequent as complements, as well as after as
or for and as appositional elaboratives, serving as descriptions of the referent
of other constituents (e.g. the subject or the object):
(24) Mr. Garg was a regular customer.
(25) They called him a damned fool.
(26) Vinod refused to see himself as a "servant".
(27) I took him for a criminal.
(28) She fell in love with Max Jones, a real-estate agent from Minnesota.
In these examples, the indefinite noun group provides a description of the
referent of some other constituent by assigning to it membership of the class
of potential referents of the head noun. It is interesting to compare such
constructions with the following examples without the indefinite article:
(29) Bill Clinton became president of the United States.
(30) As chief director of this firm I disapprove of your 'useful contacts'.
Here the italicized noun groups without the indefinite article have unique
reference rather than class-member reference.
The indefinite article is also sometimes left out to create an emphasis on
different aspects or (changing) 'character' or 'quality' of the referent described
rather than simply assigning class membership, especially in constructions
with as or after the verb TURN (cf. Schibsbye 1970: 287):
(31) As American, the writer has distrusted Europe; as writer, he has envied the
riches available to his European counterpart.
(32) He had turned spy.
Determination 375
This emphasis on 'character' or 'quality* is also found in connection with kind

of and sort o/expressions; compare:
(33) What sort of man/α man would do such a thing!
Such examples are more frequently without the indefinite article.
Note finally examples involving enough (cf. Schibsbye 1970: 287):
(34) I am still optimist enough to credit life with invincibility.
C) Indefinite specific vs. indefinite non-specific reference. As with groups
containing the definite article, groups containing the indefinite article are
susceptible to an analysis in terms of the distinction between specific and
non-specific. Most of the examples offered so far in this section are examples
of indefinite specific reference: in each case a particular referent answering
the description of the noun group is picked out specifically. Thus in example
(22) above, an elephant does not simply refer to a random member of the
class of elephants, but to a particular member. When an indefinite noun
group does not refer to a particular member but more loosely to potentially
any member of the class, we speak of indefinite non-specific reference. Here
are some examples of indefinite non-specific reference (with indefinite
specific counterparts in parentheses):
(35) I would like an apple, please, (cf. He gave me an apple)
(36) Are we likely to see a viper in this region? (cf. We saw a viper here
yesterday)
Here the indefinite article approaches the meaning of 'any': any entity
answering the description of the head (plus modifiers) is a potential referent.
Some noun groups are ambiguous between a specific and a non-specific
reading:
(37) Sally wants to marry a Norwegian who is rich.
The noun group a Norwegian who is rich either refers to a particular person,
a rich Norwegian (specific reference), or it marks anyone who is a rich
Norwegian as a potential future husband (non-specific reference).
D) The indefinite article in generic-like expressions. Moving one step
further away from indefinite specific reference, the indefinite article is used
in general statements to denote 'typical class-member1, cf. the following
examples:
(38) An elephant is a potentially very dangerous animal.
(39) A linguist is someone who studies languages.
(40) A funnel-web spider is poisonous.
376 Nominate
The indefinite article here implies 'any' but differs from any in that it does
not single out the members of the class individually but rather has a generic-
like function. Such constructions are more restricted in general statements
than truly generic plural expressions with zero article (e.g. Elephants are
potentially very dangerous animals). Thus, as Greenbaum and Quirk point
out (1990: 85), there are cases where we cannot use the indefinite singular
expression:
(41 a) Elephants are becoming extinct.
(41b) * An elephant is becoming extinct.
Example (4la) is a statement about the species as a whole and therefore
cannot be replaced by example (41b) with its class-member association.
Generic-like expressions with the indefinite article also differ from generic
expressions with the definite article (e.g. The elephant is a potentially very
dangerous animal) in being less formal and less contrastive. Note here that
generic singular expressions with the definite article express 'class as a
whole' rather than 'typical class-member' and are not subject to the restriction
mentioned above:
(41c) The elephant is becoming extinct.
£) Count nouns with little association of class-membership. The
indefinite article is sometimes used in idiomatic expressions where it does
not really make sense to speak of class-member reference, (cf. Jespersen
1933: 175):
(42) I have a mind to tell him exactly what I mean.
(43) The child was in a fever.
(44) I did it with a view to being useful.
Note especially constructions where there is fusion between the predicator
and the direct object in S P O and S P O O constructions (creating
semantically intransitive and monotransitive constructions, respectively, cf.
our discussion of role suppression in section 7.3.4):
(45) He gave a nod. (cf. He nodded)
(46) They had an argument, (cf. They argued)
F) The indefinite article and mass nouns. As a general rule, mass nouns do
not take the indefinite article. When unmodified, they typically appear
without article (generic use) or with quantifiers like SOME:
(47) Sugar is more expensive than rice.
(48) Could I have some water, please.
Determination 377
However, as pointed out in section 10.2.5, the indefinite article is used in

connection with mass nouns to denote 'subtype of the non-countable entity1
(especially in noun groups containing restrictive modification as in example
(49)) or 'instances' of the non-countable entity (as in example (50)):
(49) This is actually an excellent wine.
(50) He ordered a beer and a coffee.
If the referent is an abstract concept and this concept is restricted by the
meaning of one or more modifiers, the indefinite article is used for indefinite
expression (cf our discussion of the use of the definite article in similar
contexts in section 10.3.5); compare:
(51) She showed a loyalty towards her master which I could not match.
(52) This was Ranjit's condemning reference to Deepa, for whom he harboured a
forbidding disapproval - the kind that only Mr. Sethna might have shared.
In these examples, the meaning expressed is 'kind of...' ('kind of courage',
'kind of loyalty' and 'kind of disapproval'; note the appositional elaborative in
the last example: the kind that only ...) and is thus similar to example (49),
which expresses 'kind of wine'.
G) The indefinite article and proper nouns. The indefinite article is
occasionally used with proper nouns to denote class-membership, as in:
(53) Dr. Aziz said, 'Have you heard about Dr. Dev?1 Farrokh wondered, Which
Dr. Dev? There was a Dr. Dev who was a cardiologist, there was another
Dev who was an anesthesiologist - there are a bunch of Devs, he thought.
(54) I was struck by his dark eyes and small ears. There was no doubt about it: he
was very much a Staines.
In these examples, the referents actually belong to a class of people bearing
the name provided by the proper noun (cf. sections 10.2.4 and 10.2.5). But
proper names may also be used with the indefinite article for members of a
class of people simply sharing some quality with the bearer of the name or
being of the same standard, as in:
(55) This country has never produced a Shakespeare or a Picasso.
(56) 'That's it! That's absolutely it! Somebody like John Aubrey, who listens to
everything, wonders about everything. This university needs an Aubrey.'
The indefinite article (as well as emphatic one) is sometimes used in connec-
tion with a proper name to indicate that the bearer of the name is unknown to
the speaker or assumed by the speaker to be unknown by the listener:
(57) Is there a Sarah Mortimer staying at this hotel?
(58) In the cupboard I found a pretty volume, the work of one William Canton.
378 Nominate
Groups containing a modified proper noun preceded by the indefinite article

typically express that the bearer of the name is in a temporary (physical or
mental) state:
(59) "I hope it won't be another 50 years before we can celebrate like this again,"
joked a high-spirited Bing Crosby.
(60) It was a young Peter Simpson that I saw in the picture.
H) The indefinite article as a basis of quantification. As a marker of
singular meaning, the indefinite article lends itself to quantified expressions
involving a standard against which something is measured or counted:
(61) They made love twice a day.
(62) It costs £45 a pound.
The indefinite article here approaches each (or per) in meaning.
10.3.7. Zero determination

The zero determiner must first of all be distinguished from cases where two
or more nouns share a determiner (as in / asked for a knife and fork, the
Danes and Swedes felt humiliated, etc., where the head of the group is
realized by a compound unit). A distinction must also be drawn between
'zero determiner', which is associated with indefiniteness, and 'no
determiner1. Names take no determiner rather than zero determiner because
they have unique (and thus inherently definite) reference, as in the following
examples:
(1) Farrokh and Julia were sharing a bath together.
(2) The view of Back Bay was stunning. Martin Mills could see Malabar Hill
and Nariman Point.
Note in this connection the lack of determiner in connection with count
nouns expressing family relations: Uncle will join Mom and Dad in a
moment. This applies also to vocatives like the following: How are you this
morning, Professor? I Come on, Darling I See you there, mate.
Let us turn now to the uses of zero determination:
A) Indefinite class-member reference. Zero is regularly used in connection
with plural count nouns for indefinite specific or non-specific class-member
reference (corresponding to the use of the indefinite article plus singular
count noun):
(3) I was going to buy clothes, (cf. / was going to buy a hat)
(4) Madhu won't be working with clowns - or with elephants, (cf. Madhu won't
be working with a clown - or with an elephant)
Determination 379
(5) She wore bells on her ankles and wrists, (cf. She wore a bell on her left
ankle)
This use of plural nouns with zero determiner is compatible with modifica-
tion and/or quantification:
(6) They wore navy-blue shorts and kneesocks, too - and black shoes.
(7) Sai Baba was a patron saint of many circus performers.
(8) She dances with two beautiful peacocks.
Indefinite class-member reference with zero determiner and quantification is
found also in examples where what is basically a mass noun is used in the
plural to denote 'instances or realizations of, as in Over the years she did me
many kindnesses, Two coffees, please, etc. (cf. section 10.2.5 above).
B) Indefinite mass reference. Zero is also used in connection with mass
nouns for indefinite specific or non-specific reference to non-countable
entities and concepts:
(9) Danny poured hot water over the peas.
(10) It contained some whitish stuff- curdled milk or flour and water.
(11) The cripple looked to Dr. Daruwalla for help.
(12) Martin would be kept in perpetual darkness.
Non-specific reference to non-countable entities and concepts, as in the last
two examples, have a generic potential (cf. subsection C below).
Note that names for diseases are usually non-countable and take zero
determiner: e.g. CANCER, PNEUMONIA, CLAMYDIA, etc.:
(13) He suffers from cancer.
(14) Her child had pneumonia.
C) Generic reference. As already noted, zero is used in connection with
plural count nouns and with mass nouns for indefinite non-specific reference.
In very general contexts, i.e. when there is no explicit or implicit limitation
on the referents of the plural noun, or the mass noun, the construction may
assume a generic meaning, referring to all the members of a class:
(15) Elephants are dangerous animals.
(16) Professors like poetry.
(17) Trachoma is one of the leading causes of blindness in the world.
(18) Time is still a mystery to science.
The generic quality of these examples is a result of the lack of restriction on
the referent of the noun (the class-members or the mass) rather than the
representative value of the noun. In other words, the referent is in principle
'all-inclusive' (e.g. Elephants actually means 'all elephants'). Restriction of
380 Nominals
the noun may however occur without this resulting in specific reference:
Indian elephants are dangerous animals, Polish professors like poetry, etc.
Such examples are subgeneric. Subgeneric reference without the article is
found also when what is basically a mass noun is used in the plural to denote
'kinds of as in He imports several coffees from Africa (= 'kinds of coffee').
Unlike other count nouns MAN and WOMAN take zero article in generic
expressions, the former sometimes in the sense of 'mankind' (though this is
often avoided to escape accusations of sexism):
(19) The temperament of man, either male or female, cannot help falling down
before and worshipping ibis sacrificial note.
(20) It was man who ended the Cold War in case you didn't notice. It wasn't
weaponry, or technology, or armies or campaigns.
(21) The nineteenth century was an age where woman, not man, was sacred; and
where you could buy a thirteen-year-old girl for a few shillings.
Count nouns occasionally serve as generic terms for academic subjects,
sports and activities:
(22) Mary studied dance as well as film. (Greenbaum & Quirk 1990: 85)
(23) All he is interested in is play ing football.
D) Generic-like expressions in specific contexts. Abstract and concrete
mass nouns, as well as plural count nouns, are sometimes used in a generic-
like way in contexts that are so specific that the meaning is close to definite
specific expressions (see e.g. Vestergaard 1985: 113):
(24) (Our) sources say he was once a paid informer for the FBI.
(25) (The) traffic had to be diverted because (the) roads were flooded.
(26) (The) productivity in most industries is lower than two years ago.
Note particularly the use of certain general count nouns like CONDITIONS,
MATTERS, EVENTS, THINGS, etc., without determiner even in rather
specific contexts:
(27) Conditions were hopeless.
(28) Don't make matters worse.
E) Other uses of zero. In our discussion of the definite and the indefinite
article in sections 10.3.5 and 10.3.6, we have already noted a certain
vacillation in the choice relation between the articles and zero determination.
Let us look at some of the finer details of this choice relation:
(i) As pointed out in section 10.3.6, count nouns take zero article when they
are used as complements, as appositional elaboratives, and after as and for
for unique reference rather than class membership: Bill Clinton became
president of the United States /She was elected chairman of the Equal Rights
Determination 3 81
Committee, etc. Note further fronted complements in concessive and

conditional subclauses:
(29) But pious man that Reverend Jackson was, he was also a father whose
children needed a mother.
(30) This fault - if fault it is - is an amiable one. (Vestergaard 1985: 30)
(ii) Count nouns are sometimes used like mass nouns to emphasize the
material, character or concept of the referent rather than simply the referent
as an individual entity (cf. also section 10.2.5), in which case they take zero
article: The scrapyard was full of smashed car / He is impervious to
argument / The proposal was discussed in great detail (cf. Vestergaard 1985:
91) /She was incapable of plotting murder / Word got round that he had
already resigned. Note especially the frequent absence of determination after
kind of and sort of. What kind of car is this /Any sort of knife will do, etc.
(iii) Count nouns of the following kind are used without determination in a
great many standard expressions for typical recurring activities, events and
conditions (cf. Greenbaum & Quirk 1990: 82f):
locatives with an emphasis on the function of the place rather than its
location: e.g. be in or go to bed/church/school/prison/college·,
• transport and communication: e.g. go/travel/come by bus/train /car, send
by post/satellite;
• meals: e.g. When is dinner? Do come for lunch tomorrow;
• time expressions: summer, autumn, morning, evening; at dawn, by night,
in winter.
As many of the examples show, the lack of article is especially frequent in
preposition groups.
(iv) Count nouns are often used without determiner in compound units even
if the individual noun would require an article:
(31) As he spoke, he cupped his hand near the bony fusion of ankle and foot,
which the beggar awkwardly rested on the heel.
(32) We could say that they are brother and sister, that one looks after the other.
A similar phenomenon is observed in fixed adverbial collocations and
idiomatic expressions like: hand in hand, side by side, mile after mile, etc.
(v) Nominals serving as dependent modifiers typically take zero deter-
mination (though the superordinate noun is usually determined): these onion
choppers, other available child prostitutes. some guest workers, etc.
382 Nominate
(vi) Nouns with zero determination often occur in fixed expressions,

especially verb + noun collocations (such as catch sight of, take place, make
use of, make contact with, take pains, take care, cast anchor, etc.) and in
preposition groups (such as at ease, at pains, in mind, in proportion to, at
first glance, off balance, for sale, in need of, in style, by hand, by surprise, on
holiday, peace of mind, out of character, in writing, in support of, in fear
that, at high speed, in case, by means of, in charge of).
10.3.8. The genitive

The genitive construction is first and foremost a central definite determiner
used to express a relation between two referents: that of the genitive
construction itself and that of the head of the group. This relation is often but
by no means always one of possession, as in e.g. the old man 's hat. This
example refers to a particular hat (and is thus much closer in meaning to the
hat than to a hat). At the same time it indicates that the hat belongs in some
sense to the referent of the old man. In its central function, the genitive
construction is called the specifying genitive. Note that a noun group
containing the specifying genitive is definite even if the genitive construction
is indefinite: a young man's desperate attempt (= 'the desperate attempt of a
young man').
In writing, the genitive case is formed by adding 's to singular nouns or
(certain indefinite) pronouns and to irregular plural nouns that do not end in
-s: father's, Jack's; somebody's, nobody's; women's, children's. The genitive
of plural nouns ending in -s is formed by simply adding the apostrophe:
boys', writers'. In speech, the genitive marker is Λζ/ after sibilants, /s/ after
voiceless sounds other than sibilants and Izl elsewhere, and: witch's /wrtj'iz/,
wife's /waifs/, boy's /boiz/. In speech, the regular plural is not marked for the
genitive: boys' /boiz/, writers' /raitaz/. There are few exceptions to these
rules: foreign, especially Greek names of more than one syllable ending in
-es take only the apostrophe: Socrates', Xerxes'. With names, English as well
as foreign, ending in -s pronounced Izl, usage varies: Lyons' or Lyons's,
/laianz/ or /laianziz/. The bare apostrophe is usual after a sibilant inôr ...
sake expressions: for goodness' sake.
Though the genitive marker is added simply to nouns (or certain
pronouns), the genitive construction potentially contains a whole noun group
or compound unit: cf. women's (disinclination) vs. the old women's (dis-
inclination) and Jack's (humour) vs. Jack and Jill's (humour).
Sometimes the genitive marker is not added to the head of the group
contained in the genitive construction but to the postmodification, in which
case we have what is traditionally referred to as a 'group genitive':
Determination 383
the King of Sweden's (decision)

somebody else's (friend)
the girl who lives next door's (bike)
The term 'group genitive' is somewhat misleading if restricted to such
examples: the genitive in cases like the old women's disinclination is also,
strictly speaking, a group genitive (the old women being a noun group as
much as the King of Sweden).
There is an obvious relationship between genitive constructions and
possessive pronouns: compare the old man's hat and his hat. By and large
possessive pronouns are used like genitive constructions. Thus much of what
we have to say about the use of genitive constructions applies also to
possessive pronouns (for some of the differences see section 11.2.2). Let us
turn now to some of the important features of the genitive.
A) Syntactic functions of the genitive. Although the genitive construction is
basically a central definite determiner (the specifying genitive), it has other
important uses. It may serve as a premodifier, entering a compound-like
relationship with the head of the group:
(1) She lives in a quaint old shepherd's cottage, (not 'the cottage of a quaint old
shepherd' but 'a quaint old cottage of a certain kind')
(2) He took a doctor's degree, (i.e. 'a special kind of degree1)
(3) She bought some children 's shoes, (i.e. 'some shoes for children')
Compare the two possible interpretations of These children's shoes are very
expensive from a sentence analytic point of view. Here is first the
interpretation of the example in terms of a specifying genitive ("shoes
belonging to these children'):
STA
expensive
shoes
These children's
384 Nominals
By contrast, the following tree diagram shows the interpretation of the

example in terms of a premodifying genitive ('these shoes for children'):
STA
expensive
These
children's shoes
The premodifying genitive, which is called the classifying genitive, differs

from the specifying genitive phonologically in having unitary stress on its
first element and syntactically in being inseparable from the head of the
group. Semantically the classifying genitive denotes 'kind' or 'type'. Note that
we cannot insert e.g. an adjective between the genitive and the head and still
preserve the status of the genitive as a classifying genitive: cf. some
children's expensive shoes I some expensive children's shoes. And while
groups containing a specifying genitive always have definite reference,
groups containing a classifying genitive may have indefinite reference (as in
the examples above). A specifying and a classifying genitive may co-occur,
as in:
(4) It was the child's first doctor's visit.
(5) It was the doctor's second doctor's degree.
Consider next what may be referred to as the autonomous genitive. The
autonomous genitive is a specifying genitive construction which does not
relate to an overt head but rather by itself assumes an external function like
that of a group:
(6) I parked my car behind Jack's.
(7) Jennifer's hat is more expensive than her mother's.
(8) George's is the only voice she is likely to recognize.
In these examples, the head is ellipted because it is contextually clear (Jack's
= Jack's car, her mother's = her mother's hat, George's - George's voice),
cf. section 4.2.2. The autonomous genitive also occurs in expressions
referring to homes, buildings, institutions, businesses and other places:
Determination 385
(9) I met her at my uncle's.

(10) I got these rolls at the baker's.
(11) He has been to London often but never actually seen St Paul's.
In such examples of what is often referred to as the 'local genitive', the
missing head may occasionally be uniquely retrievable (as in the case of St
Paul's Cathedran: as a rule, however, the missing head could be any one of a
number of nouns belonging to the same semantic type, viz. 'locations' (my
uncle's = my uncle's place/home/residence/etc.).
Finally the autonomous genitive may appear in a postmodifying of-
construction with a quantifying, partitive meaning (Of several'). This genitive
we call the post-genitive; examples:
(12) He introduced me to some friends of my neighbour's.
(13) Jack borrowed a picture of my sister's.
The post-genitive, which provides an indefinite alternative to the specifying
genitive (cf. my neighbour's friends, his sister's raincoat), is compatible with
a head noun determined by the definite article or a demonstrative pronoun
when followed by a restrictive relative clause: the/that/this unfortunate
student of Otto Jesper sen's who failed the grammar exam. The partitive
content becomes especially clear when we consider similar constructions
with the uninflected noun:
(14) Jack borrowed a picture of my sister.
(15) He was a student of Otto Jespersen.
(16) He introduced me to some friends of my neighbour.
There is in these examples no partitive meaning but rather a relation defined
by the head of the group directed towards the complement of of (i.e. 'Jack
borrowed a picture representing my sister1, 'he studied Otto Jespersen',' he
introduced me to some people who consider themselves to be friends').
Nouns incapable of expressing such a unidirectional relation do not appear in
this kind of construction: *He wore a raincoat of my sister.
In examples containing demonstrative pronouns like:
(17) Now tell me something about that/this brother of Stephanie's.
the partitive association is replaced by an association of the speaker's
(approving or disapproving) interest in the referent.
Here is an overview of the syntactic functions of the genitive:
386 Nominate
central determiner the old man's hat

(specifying genitive)
premodifier α ladies' magazine

Genitive
(classifying genitive)
elliptical genitive I parked my car

behind John's
without group head local genitive at my uncle's

(autonomous genitive)
post-genitive a good friend of my

neighbour's
The rest of this section is devoted to the specifying genitive.

B) The semantics of the specifying genitive. The specifying genitive is
traditionally said to express possession in a broad sense: the doctor's right
hand, his wife's only other reading material, another family's mansion, My
son-in-law's cigarettes. In an example like the girl's bathtub at the Hotel
Bardez, the bathtub is more likely to belong to the hotel than to the girl; and
in John's bike, John may or may not be the owner of the bike (for example,
the bike could simply be one that John is riding).
The fact is that in many cases, there is simply just some unspecified
relation between the referent of the genitive construction and the referent of
the group as a whole. This has led to various classifications of the specifying
genitive on a semantic basis. Apart from the possessive genitive, we get e.g.
the genitive of origin (as in Jack's letter 'the letter is from Jack'), the genitive
of attribute (as in Nancy's irritation 'Nancy is irritated'), etc. Two distinct
kinds of genitive have attracted special attention: the so-called subjective
genitive and the so-called objective genitive. These genitives are relevant in
connection with verbal nouns, such as ESTIMATION, DELIVERY, EM-
BRACE, RIDICULE, etc., to which we can assign potential participant roles
usually associated with syntactic 'subject' and/or Object' function.
In the following examples, the specifying genitive is a subjective genitive
in the sense that the referent of the genitive construction is the DOER of an
(explicit or implicit) situational referent of the group as a whole:
(18) In Dr. Daruwalla's estimation, the Jesuits were intellectually crafty and sly.
(19) Patel's cool delivery of the news appalled me.
Determination 387
(20) Dr. Daruwalla sat shivering in Julia's embrace.

(21) Julia wasn't amused by Farrokh 's constant ridicule of St. Francis 's violated
remains.
When the head of the group is a nonfinite clause, the referent of the
subjective genitive is either the active DOER/SPECIFIED or the passive
DONE-TO of the situation expressed by the clause:
(22) Jenny's neglecting to write that letter upset John. (Jenny = DOER)
(23) Jack's being so considerate lately made her suspicious. (Jack = SPECIFIED)
(24) Jenny's getting pushed over the edge was an accident. (Jenny = DONE-TO)
While the subjective genitive is frequent, the objective genitive has rather
restricted occurrence, the objective relation being often expressed more
naturally by an o/construction:
(25) Was there another clue to Mr. Lai's murder, or another threat to Dhar? (= "the
murder of Mr. Lai1)
(26) She relished the details of old Jack's release. (= 'the release of Jack')
(27) Without this evidence St. Francis's canonization might never have occurred.
(= 'the canonization of St. Francis')
In these examples, the referent of the genitive construction is semantically
the DONE-TO.
Note that if both a subjective and an objective relation are to be expressed,
we have a subjective genitive followed by an o/construction expressing the
objective relation: Jack's release of the prisoners (= 'Jack released the
prisoners'). Objective genitives may, however, be accompanied by a follow-
ing by-phrase expressing the subjective relation: Denmark's occupation by
Germany (cf. the occupation of Denmark by Germany).
In connection with head nouns derived from intransitive verbs, the o/-
construction unambiguously expresses the subjective relation: cf. the king's
arrival / the arrival of the king.
C) The specifying genitive vs. the ^/-construction. The use of the genitive
is somewhat restricted in the sense that many nouns do not normally occur in
the genitive: it would be fairly unusual to say e.g. the table's price, the
house's roof, obedience's ramifications, his conversion's miracle, etc. To
express the intended meaning, we often use an entirely different syntactic
construction: the definite article + head noun + an o/construction containing
the referent to which the head is related (the price of the table, the roof of the
house, the ramifications of obedience, the miracle of his conversion). In
many cases, both expressions are possible with little or no difference of
meaning: Veronica Rose's offspring / the offspring of Veronica Rose, the
girl's mutilated body / the mutilated body of the girl, etc. For this reason,
388 Nominate
traditional grammar used the term O/-genitive' for this construction and
provided rules for the use of the 'two competing genitives'. We do not use the
term 'genitive' about the o/-construction because, syntactically, it is com-
pletely unrelated to the genitive: the genitive is a determiner while the o/-
construction is a postmodifier. But since semantically the two kinds of
construction are often choice-related, we do find it relevant to provide some
guidelines for their use. Let us review some of the factors determining the
choice of construction, keeping in mind that there are few absolute rules
involved:
a) Semantic considerations. The greater an association of humanness,
animacy and/or individuality, the more likely we are to be able to use a noun
in the genitive. Thus names of persons and higher animals and count nouns
referring to people usually take the genitive (e.g. Jim's book, Fido's kennel;
the girl's arm, the teacher's car, etc.) while nouns referring to inanimate
entities usually take the o/-construction (e.g. the colour of the wall, the other
side of the coin, the result of this test, etc.).
Geographical names, collective nouns, count nouns referring to institu-
tions, regions, places, etc. are common in both constructions: Goa's white
beaches /the white beaches ofGoa; my family's reputation / the reputation of
my family, the clinic's name /the name of the clinic', etc.
It is often claimed that more 'individuality' or 'focus of interest' is placed on
an inanimate or non-human referent of a noun by using it in the genitive
rather than in the o/-construction in examples like the bikini's bottom half, a
lizard's eyes, the novel's title, the fire's friendly crackling, the envelope's
shape and size, etc. In practice such distinctions are hard to perceive.
Abstracts rarely take the genitive (the significance of this concept, the
beauty of this idea, much rather than this concept's significance, this idea's
beauty) unless personified or individualized (nature's wonderful solution to
that problem, life's many mysteries).
As already noted above, the objective genitive is rarer than the possessive
and the subjective genitive, being typical only in connection with names and
nouns referring to persons (John's defeat, the woman's release}.
Time and distance expressions often appear in the genitive: within two or
three weeks' time, at a yard's distance, a moment's reflection, etc.
When a noun relates to one of the lexical items EDGE, END, SURFACE and
SAKE as head of the noun group, we often get a genitive construction rather
than an o/-construction: the water's edge, the journey's end, the lake's
surface,for brevity's sake, etc. In the case of WORTH only the genitive
construction is possible: his money's worth.
Determination 389
Note finally idiomatic expressions like: Not for the death ofme / Not for the
life of me /I don't like the look ofthat man.
b) Formal considerations. Strings of genitives are often avoided (Martin 's
heart's desire, my cousin's wife's first husband). Strings of o/-constructions
are common even if somewhat clumsy (many of the conclusions of the
report). Mixtures of the two constructions are often a happy compromise (the
first husband of my cousin's wife, many of the report's conclusions), cf.
Schibsbye 1970: 117.
Heavy genitive constructions are generally avoided in formal language,
especially if the genitive marker can only be placed on a postmodifier (cf. the
man I met yesterday's wife / the wife of the man I met yesterday, the former
division officer in this firm's secretary / the secretary of the former division
officer in this firm). Considerations of end-focus and end-weight are
important in cases like the former prime minister's daughter vs. the daughter
of the former prime minister.
O/~-constructions are used in order to make it possible to attach a relative
clause or participial construction directly to the head it modifies, cf. the
following example from Vestergaard 1985: 107:
(28) There is a characteristic 'double deprivation1 in the lives of these children,
who tend to deprive themselves further through ... (.. in these children's lives,
who tend to..)
Often we get a combination of considerations, e.g. heaviness, postmodi-
fication and avoidance of the plural genitive, as in the following example:
(29) These were usually the houses of producers, directors or actors to whom
Danny owed a finished screenplay.
Adjectives used with generic nominal referents, as in the poor, the rich, the
merely fanciful, etc.) do not take the genitive case: * the poor's conditions /
the conditions of the poor, cf. section 12.4.1.
When the head of the group is realized by a nonfmite clause, only the
genitive is possible: Jenny's neglecting to write that letter / *the neglecting to
write that letter of Jenny.
In cases where the genitive singular and plural are identical in sound
(friend's/friends', girl's/girls', etc.), the »/-construction is sometimes used
instead of the genitive plural to avoid ambiguity: his friend's opinion / the
opinion of his friends. In writing, such plural genitives are common:
(30) He awakened to the sound of the skiers' boots tramping on the hard-packed
snow.
(31) It was at least partly the passion of such writers' convictions that gave their
novels such value.
390 Nominals
10.4. Quantification: the number category

As already pointed out in sections 10.1.3, 10.2.5 and 10.3.3, quantification is
one of the central subfunctions within the functional domain of nominals: we
rarely code specific things without quantifying them, either implicitly or
explicitly. One of the most explicit ways of quantifying something is by
using exact terms of measurement, e.g. cardinal numbers: six apples, two
bottles of apple juice, etc. But quantification is not limited to expression by
such lexical means. Nor is it tied down to one specific component in the
noun group (such as cardinal numbers). It rather involves a number of
different types of constituent: the head noun (with its singular/plural number
contrast: girl/girls, car/cars, etc.), certain determiners (which show number
concord with the head: the indefinite article (a), indefinite pronouns (such as
every, each, (n)either) and demonstratives (that/those, this/these)), and
quantifiers serving as predeterminers, postdeterminers or modifiers (such as
all, both, many, five, numerous, few, etc.). In the following subsections we
examine a central grammatical means of quantification: the singular/plural
number distinction in nouns.
10.4.1. The regular singular/plural distinction

Number is a morphological category in English with the singular as the
unmarked base form and the plural as the morphologically marked form. In
speech, the plural is formed by adding:
(i) /iz/ to nouns ending in sibilants: e.g. noses, niches, fringes
(ii) /s/ to nouns ending in voiceless non-sibilants: e.g. cats, kicks, taps
(iii) lit to other nouns: e.g. dogs, clans, shows
In writing, the plural is usually formed by adding 's1 to the base form (as in
the examples above). Note however the following modifications to this
simple spelling rule:
(i) -s or -es. Nouns ending in a sibilant take -es instead of simply -s unless
they are written with a silent -e: latch/latches, mass/masses, etc. (but e.g.
bridge/bridges).
(ii) Consonant doubling. There is doubling of the final consonant mfezzes
and quizzes. Occasionally bus is pluralized as busses instead of the more
regular buses. Note also that a number of abbreviations are pluralized by
means of consonant doubling alone: e.g. pp. 1-5 (= 'pages 1-5'), exx. (=
'examples'), MSS (= 'manuscripts').
Quantification: the number category 391
(iii) -y —> -ies or -ys. The plural form of nouns ending in written -y is -ies (as
in fly/flies, cry/cries, ally/allies, etc.) except if -y is immediately preceded by
another written vowel (as in toys, delays, ways, etc.). Exceptions to this rule
are: most proper nouns (Marys, Germanys, Julys', but the two Sicilies) and
members of other word classes (stand-bys, the whys and wherefores). Note
also soliloquies, where -qu- is regarded as a consonant group.
(iv) -o —» -oes or -os. Some nouns ending in -o take plural -es: echoes,
heroes, potatoes, tomatoes, vetoes. Others (especially proper names, abbre-
viations and cases where -o is immediately preceded by another written
vowel) take plural -s only: Eskimos, Neros; kilos, photos, pianos', embryos,
studios; etc. There is vacillation in: banjo(e)s, buffalo(e)s, cargo(e)s,
commando(e)s, halo(e)s, motto(e)s, volcano(e)s, and others. Note the
difference between bravos (= applause) and bravoes (= bandits).
(v) -s or -'s. The plural form of letters is -'s rather than simply -s: e.g. p's and
q's. With abbreviations and numerals written in figures there is vacillation
with -s as the commoner form: MP's or MPs, 1980's or 1980s. Quoted words
usually take -'5: There were too many hut's in the passage. Words which do
not merely function as quotes but assume an integrated meaning in the
sentence take -s: Some of his whys are hard to answer (Schibsbye 1970: 94).
Pluralization sometimes results in a sound change:
(i) /-Θ/ -» /-oz/. This change is very common in nouns like baths, mouths,
paths, youths. There is vacillation (/-9s/ or /-oz/) in oaths, sheaths, truths,
wreaths. If -th is preceded by a consonant, or a short vowel, or a written -r-
we only get /-9s/: healths, lengths; deaths; moths; births, hearths.
(ii) /-s/ —> /-ziz/. This change affects only one word: houses.
(iii) /-{/ —» l-\zl. This change affects both the pronunciation and the spelling
of nouns like calf/calves, half/halves, knife/knives, leaf/leaves, life/lives,
loaf/loaves, self/selves, shelf/shelves, thief/thieves, wife/wives, wolf/wolves.
However, most nouns ending in /-f/ take plural /-fs/: beliefs, chiefs, cliffs,
coughs, cuffs, flagstaff s, laughs, paragraphs, roofs, sniffs. There is vacil-
lation in: dwarfs/dwarves, hoofs/hooves, scarfs/scarves, wharf/wharves. Note
the difference between staffs (= groups of people working together) and
staves (= sticks, rods). Note also the painting term still lifes.
10.4.2. Irregular plurals

By far the majority of English nouns follow the reasonably regular pattern of
pluralization described in section 10.4.1 above. However, there are a number
of important exceptions:
392 Nominate
(i) Vowel change. This form of pluralization is found in the following nouns:
man/men, woman/women, foot/feet, goose/geese, louse/lice, mouse/mice,
tooth/teeth. Correspondingly, we find a vowel change in compounds
containing these nouns (e.g. dormouse/dormice, gentleman/gentlemen,
chairwoman/chairwomen). Note that there is normally no difference in
pronunciation in compounds containing -man: both the singular and the
plural is pronounced [man]. Words containing -man which are not
compounds take the regular plural: Normans, Germans.
(ii) -en/-ren plural Only three nouns take the -en/-ren plural ending, two of
them with additional vowel change: child/children, ox/oxen and brother/
brethren (used only about the members of a religious community; brothers is
the normal plural of brother).
(iii) /-s/ (-ce) after voiced sound. This plural ending is found in two words:
dice (the corresponding singular die is used only in standard phrases like The
die is cast) and pence (used about amounts and 'small change', in contrast to
pennies, which is used about the individual coins).
(iv) Foreign plurals. Many nouns of foreign, especially Latin or Greek,
origin take foreign plural forms rather than the regular English plural form,
though in many cases there is vacillation with the regular form as the less
formal. Here are some of the most common examples:
• -us —> -i (/ai/) or -a (/a/)·
alumnus/alumni, stimulus/stimuli; cactus/cacti or cactuses, octopus/octopi or
octopuses. Note regular examples like: campus/campuses, genius/geniuses,
virus/viruses. Corpus and genus take the irregular plural form corpora and
genera in formal, technical language.
• -a —> -ae (/-i:/):
alumna/alumnae, larva/larvae, vertebra/vertebrae. Note regular examples
like: area/areas, villa/villas. There is vacillation in antenna/antennae (of
insects) or antennas (= aerials), formula/formulae (mathematical formulae)
or formulas (more generally).
• -um —» -a (/a/)·
addendum/addenda, bacterium/bacteria, erratum/errata and others. Note
regular album/albums, museum/museums, asylum/asylums. Vacillation:
aquarium/aquariums or aquaria, symposium/symposiums or symposia. Note
that the plural form of datum is, strictly speaking, data, but data is in-
creasingly used as a singular mass noun instead of datum, especially in
scientific language, and especially about 'a collection of facts, examples, etc.'
• -ex or -ix —> -ices:

codex/codices, index/indices (in science) or indexes (more generally),
appendix/appendices (in books) or appendixes (in anatomy), matrix/matrices
or matrixes.
• -is —> es (/-i:z/):
analysis/analyses, axis/axes, basis/bases (the plural spelling thus coinciding
with the spelling of base/bases), crisis/crises, diagnosis/diagnoses, hypo-
thesis/hypotheses, etc. The regular English plural is rare: metropolis/
metropolises.
• -on —> -a (/a/):
criterion/criteria, phenomenon/phenomena but demon/demons, electron/
electrons, etc.
• Others:
portmanteau/portmanteaus or portmanteaux, bureau/bureaus or bureaux,
corps (fl£3i/)/corps (/lo:z/).
(v) Identity of form. With certain count nouns there is no formal difference
between the singular and the plural. This applies to:
• some animal names (deer, grouse, plaice, salmon, sheep, snipe, trout)',
others vacillate (buffalo(es), antelope(sj);
• nationality names in -ese (Chinese, Japanese, Portuguese) and Swiss;
• craft (= 'boat') and compounds containing -craft (aircraft, spacecraft);
counsel; offspring;
• the following special nouns denoting number, weight or measure: head
(as in five head of cattle), brace, yoke, gross, horsepower, hundredweight,
stone).
There is also no formal distinction between the singular and the plural of the
following nouns ending in -s: barracks, gallows, headquarters, innings,
means (= 'method(s)'), series, species, works (and compounds containing
-works: gasworks, waterworks). When means means 'money' it is always
plural. Unlike the nouns mentioned under a) to d), these -s nouns may be
used as plurals (e.g. with respect to concord) even if they refer to a single
entity (van Ek and Robat 1984: 101): Our old headquarters were abandoned.
394 Nominate
10.4.3. Number-invariable nouns

Some nouns are invariably either singular or plural, or at least restricted with
respect to the singular/plural contrast. For example, as we have seen, mass
nouns (like remorse and mud) are singular (unless they express 'instances' or
'kinds of as in e.g. kindnesses and wines). The same applies to situation-
referring -ing forms: cf. Their dancing on the table is unlikely to please him
vs. His writings have caused quite a scandal. Names are either singular (such
as John, Denmark, the University of Sussex) or, less often, plural (such as the
Hebrides, the Alps, General Motors), but they are usually invariable with
respect to quantification (for the use of proper nouns for class-member
reference in examples like the two Peters, see sections 10.2.4 and 10.3.5).
Let us look a some more specific cases involving number-invariable nouns:
(i) Nouns with 'plural form' used as singular nouns. Several kinds of noun
ending in -s fall under this heading:
• Nouns ending in -ics denoting a science or a subject: acoustics, athletics,
mathematics (also maths), linguistics,pragmatics, etc. (e.g. Pragmatics is_ a
fascinating new subbranch of linguistics). Note that politics may be treated
as a plural noun if it denotes an individual's views (His politics were
becoming an embarrassment to the government). Similarly, statistics can be
used more loosely about 'figures', in which case it contrasts with singular
statistic.
• Nouns ending in -s denoting diseases are usually used as singular nouns:
mumps, shingles, rickets, measles.
• Nouns ending in -s denoting games: billiards, cards, darts, dominoes, etc.
• News is always singular: No news is good news, This news is very
depressing. The same applies to shambles: The house is in a shambles.
(ii) nouns with 'singular form* used as plural nouns. Under this heading
we find aggregate nouns (i.e. nouns referring to composite entities) like
cattle, clergy, police, poultry, vermin, people, crew, staff. These nouns
behave differently with respect to quantification:
• Cattle is compatible with large and/or imprecise numbers: There were
fifty/some cattle in the field, He bought hundreds of/1,000 cattle (but ?I saw
three cattle is unusual).
• Clergy and police accept precise quantification even in small numbers:
There were three clergy and 12 police present at the meeting.
• Poultry and vermin do not usually allow of precise quantification: a lot of
poultry/vermin; *seven /*a hundred poultry/vermin.
• People, crew and staff are 'internally countable' like police: ten/several
people/crew/staff. Unlike police, however, they can be used as singular
nouns with regular plural forms: e.g. The Danes are a tough people/There
are several English-speaking peoples.
(iii) nouns with 'plural form1 used as plural nouns. We here include the
following:
• The so-called binary nouns (nouns referring to entities which consist of
two equal parts, usually instruments or articles of dress): binoculars, glasses,
forceps, scissors, pliers, pincers, scales; jeans, pants, trousers, slacks, tights.
To indicate a number distinction, we here have to use partitive constructions
like a pair of/several pairs of glasses/scissors/jeans/etc.
• A number of nouns which regularly occur in the plural form with a
meaning that has no obvious counterpart in the singular: airs (as in to put on
airs), brains (He has got brains), contents (= that which is contained, as in
table of contents), colours (as in e.g. to join the colours), customs (= import
duties), fireworks, funds, goods, greens, looks (as in his good looks), media,
oats, odds, outskirts, pains (= 'care', as in to take pains), premises (=
building, location), remains, riches, savings, spirits (= strong liquour), stairs,
surroundings, thanks, wages. Some of these are found in the singular but
then usually with a different meaning: e.g. content (= 'that which is written or
spoken', or about 'proportion', as in the silver content of this spoon), custom
(= habit), pain (= physical suffering), premise (= hypothesis, part of a formal
argument), spirit (= mind, soul); these singular nouns have regular plural
forms which preserve the meaning of the singular form.
Obviously related to this last group are the so-called 'intensive plurals', i.e.
plurals with a distinct meaning which is, however, related to the concept of
the corresponding singular mass or count noun: apologies (as in She sent her
apologies), fears (She felt grave fears for him), gardens (e.g. the botanical
gardens), orders (She was under orders to kill her boss), regrets (e.g. He
expressed his regrets), sands (area of sand), skies (the sunny skies of Italy),
waters (the waters of the lake, the stormy waters of the Atlantic), etc.
10.4.4. Collective nouns

As pointed out in section 7.6.4, collective nouns are count nouns which in
BrE may serve, in unchanged form, as either singular or plural depending on
the speaker's view of the referent as either a unit (singular association) or a
collection of individuals or individual entities making up a unit (plural
association): audience, board, committee, council, family,ßock, government,
party, team, etc.:
396 Nominate
(1) My family disagree about almost everything.

(2) My family disagrees with her family about almost everything.
(3) The audience was definitely hostile, (cf. van Ek & Robat 1984: 108)
(4) The audience were running to the emergency exits, (cf. van Ek & Robat
1984: 108)
Unlike aggregate nouns like people, crew and staff (see section 10.4.3),
collective nouns are not 'internally countable': we cannot say *There were
three family and four committee present at the meeting. On the other hand,
like people, crew and staff, singular collective nouns may be pluralized to
denote more than one unit: families, governments, etc.
10.4.5. What is pluralized?

Usually there is no problem in identifying the relevant unit to pluralize in a
noun group: it is, of course, the head noun. Thus we say e.g. the two chaps
from London, not the two chap from Londons. In other words, there is
normally no such thing as 'group plurals' (cf. our discussion of the group
genitive in section 10.3.8). In compound units with nouns as conjoints, each
noun is normally pluralized if plural meaning is intended: e.g. knives and
forks.
With compounds and noun + noun combinations, the plural suffix is
usually added to the last element if it is a noun: headmasters, toothpicks, city
halls, state universities, etc. Even if the final element is not a noun, we
pluralize the last element if the compound is felt to be a regular unit:
bucketfuls, breakdowns, stowaways, knock-outs, break-ins.
The first element of a hyphenated compound is pluralized in the following
cases:
(i) Noun + preposition + noun combinations: fathers-in-law, men-of-war,
commanders-in-chief, etc.
(ii) Noun + adverb combinations if the noun expresses the agent: lookers-on,
passers-by, runners-up. Note, however, also goings-on.
(iii) Noun + adjective combinations: postmasters-general, poets-laureate,
etc. but there is often vacillation: courts-martial/court-martials, attorneys-
general/attorney-generals.
Both elements in noun + noun combinations are pluralized in the following
cases:
(i) Compounds where the first noun is woman or man (also gentleman)
denoting the sex of the referent: men-servants, women doctors, gentlemen
thieves (such examples contrast with man-eaters, woman-haters, etc.).
(ii) Certain formal titles: Knights Templars, Lords Chancellors, Lords

Justices.
There is vacillation in combinations consisting of title + name (cf. Schibsbye
1970: 99): the two Miss Smiths / the two Misses Smith, the latter being the
more formal variant.
10.4.6. The uses of the singular and the plural

The number category is primarily used to make a distinction between One'
and 'more than one' with regard to the referent(s) of count nouns (or nouns
used as count nouns). However, as we have already seen, matters are
considerably more complex (see especially section 10.4.3). In this section we
shall briefly mention a number of specific rules for the use of the singular
and the plural, which do not follow automatically from any general
characterization of the number distinction.
(i) The distributive plural. The distributive plural is common in con-
structions consisting of two or more premodifying adjectives and a head
noun where the adjectives refer to separate entities, the head noun being
often pluralized even if each adjective relates to a single entity only: the
sixteenth and seventeenth centuries, the English and French nations, the
third and fourth chapters, etc. (cf. Jespersen 1933: 204). The singular is
sometimes used to avoid ambiguity, cf. in this and the following chapter vs.
in this and the following chapters.
The distributive plural is also found in constructions consisting of head
noun + of+ compound unit, as in the reigns of Elizabeth and James, the ages
of 14 and 18. Similarly in reciprocal expressions, especially with verbs like
CHANGE, SWAP, etc.: They changed seats, He changed trains at Reading,
The two women swapped husbands, They shook hands. For discussion of
individualizing expressions like Bob and John took their hats off (i.e. 'each
his own hat'), They ought to be having their bottoms kicked and their noses
tweaked (Schibsbye 1970: 107), see section 7.6 on concord.
(ii) Noun groups as modifiers. When noun groups function as modifiers or
as parts of modifiers, the head noun is singular even if it would have been
plural in the corresponding independent noun group; compare: three pieces /
three-piece suits; four courses / a four-course dinner, eight hours / the eight-
hour day; ten years / a ten-year-old boy; six feet / a six foot tall boy, etc.
There is occasionally vacillation with single nouns serving as modifiers or
first elements of compounds (Schibsbye 1970: 106): a wage(s) agreement,
trouser(s) pocket. The plural is used when it has a distinct meaning (cf.
398 Nominate
section 10.4.3) and thus serves to avoid misunderstanding: a customs officer,

a goods train.
(iii) Number and weight. The numbers dozen, score, hundred, thousand,
million and billion are not pluralized when they are (part of) dependents or
when, as heads, they are preceded by definite numerals (cardinal numbers):
(1) two hundred bikes / *two hundreds bikes (but hundreds of bikes / *two
hundreds of bikes)
(2) a few thousand cars / *a few thousands cars (but several thousands of cars /
*four thousands of cars)
(3) How many bikes were there? - About two hundred / *two hundreds
(4) Can you count to four thousand / *four thousands?
Foot ana pound remain singular when followed by a numeral: four foot two,
five pound fifty. Both the singular and the plural of ton is found when it
serves as the head of the group: two ton(s) of flour.
(iv) Collectivizing. Many animal names have both singular and plural forms
(e.g. lion/lions, elephant/elephants, duck/ducks, etc. When regarded as food
or as prey, the singular is used with a collectivizing effect (Huddleston 1984:
240; van Ek & Robat 1984: 1010; compare: They have shot several lion / We
saw several lions in the park; We bagged three elephant that day / Three ele-
phants came running towards us.
11. Pronominals
Like nominals, pronominale (i.e. pronouns or pronoun groups) may be used

to express the participants involved in situations. But while nominal
constituents serve this communicative function by means of a strong lexical
element (the noun) categorizing some entity, pronominal constituents do it
without specific categorization, either deictically by determining the referent
directly in relation to the communicative situation (e.g. / and you, which
represent the speaker and hearer of the utterance, respectively) or more
indirectly by representing referents already established by nominals in the
linguistic context (e.g. When Roger finally got hold of Rebecca, he. did not
even tell her about the deal). In addition, as we saw in section 10.3.1,
pronominals often serve as determiners (as in e.g. her wish). In this chapter
we examine the various types of pronominal with a view to establishing their
functional domain more precisely.
11.1. Preliminaries
11.1.1. Definition of the pronoun group
As will be recalled, a pronoun group is defined as a group with a pronoun as
head (cf. section 3.3.1). Examples:
(1) This must be someone new.
(2) Which of them called out?
Pronoun groups are relatively rare. One reason for this is that pronouns do
not specifically categorize referents but instead single them out directly in
the communicative context (i.e. deictically) or more indirectly in the
linguistic context (textually). In the latter case, they easily represent whole
noun groups (not just nouns, as the term 'pronoun' may lead one to believe),
and therefore do not often require group status themselves:
(3) My little sister thinks she is the boss around here.
(4) The old gentleman staying in the room at the end of the corridor hardly
recognized himself in the mirror.
In these examples, she does not simply represent the head noun sister but the
whole subject group My little sister, and A/mse//r'represents not only the head
noun gentleman but the whole subject group The old gentleman staying in
the room at the end of the corridor.
The main emphasis of this chapter is therefore on single pronouns.
400 Pronominals
11.1.2. Classification of pronouns

There are several distinct types of pronoun:
personal: I/me, you, he/him, she/her, it, we/us, they/them;
possessive: my/mine, your/yours, his, her/hers, its, our/ours,
their/theirs;
reflexive: myself, yourself, himself, herself, itself, ourselves,
themselves;
demonstrative: this/these, that/those;
interrogative: who/whom, which, what; whoever, whichever,
whatever; where, when, how, why;
relative: who/whom, which, what, that, 0; whoever,
whichever, whatever; where, when, why;
indefinite: any/anybody/anyone/anything
every/everybody/everyone/everything
no/nobody/no one/none/nothing
some/somebody/someone/something
all, each, both, either/neither, one/ones, other(s).
Pronouns always have determinative force: definite (personal, possessive,
reflexive, demonstrative), interrogative, relative or indefinite (cf. section
10.3.1).
Personal, possessive and reflexive pronouns are sometimes grouped
together as 'central pronouns': they are special in showing person distinctions
and in being formally related (e.g. possessive pronouns can be regarded as an
inflectional case variant - the genitive - of personal pronouns).
Interrogative and relative where, when and why, and interrogative how, are
traditionally classified as adverbs rather than pronouns. However, as the term
'pronoun' is often used more generally about pro-forms, it seems reasonable
to include them in this chapter. As pro-forms they typically represent
preposition groups (e.g. 'at what place', 'for what reason', etc.).
Even on a strict delimitation of the class of pronouns, it would not be easy
to define pronouns rigidly. A number of items seem to serve pronoun-like
functions. Among them are quantifiers like much, many, more, most, little,
less, least, few, several, half, enough, other(s) and another, ordinals (e.g.
first, second, etc.), cardinals (e.g. two, three, etc.), so, such, same, demon-
strative expressions like the former and the latter, the adverbial pro-forms
here, there, now and then and the provisional subject there, which may all be
used representatively like pronouns:
Preliminaries 401
1l) Many regard him as a genius.

(2) The two never returned.
(3) Such was her beauty that I was speechless, and so was my brother.
(4) There were many students in the hall.
In these examples the representative function of the italicized constituents is
fairly obvious (for example, in (1) Many represents e.g. "many people1). But
are they pronouns?
A possible way of distinguishing genuine pronouns from members of other
word classes is to see if, as group heads, their determinative force is such that
they are incompatible with the definite and indefinite article. According to
this criterion, Such and so in (3) are pronominal while two in (2) is not. More
generally, this criterion allows us to exclude ordinals, cardinals and adjec-
tives used nominally (as in the rich, the unthinkable, etc., see section 12.4).
Unfortunately, however, some items traditionally treated as pronouns are
also excluded: one and others (cf. She gave me a new one, The others may
refuse). Few, many, more, most and less are borderline cases: though they do
not generally accept the definite or indefinite article, there are exceptions,
such as the idiomatic or fixed expressions the chosen few, a good many, the
more the better, I like her the most, etc. Note also the distinction between
few/little and a few/a little (cf. section 7.5.5). Many ana few only accept the
definite article regularly if restrictively postmodified, as in The many who
wanted to leave and the few who are likely to support you.
We conclude that it is difficult to delimit the class of pronouns rigidly from
other word classes.
11.1.3. The external relations of pronominals

Initially it is convenient to distinguish two main syntactic uses of pronouns:
autonomous and determinative. Some pronouns only have autonomous use
(e.g. he, somebody, there, who, it, as in He called somebody}. A few only
have determinative use (my, your, their, no, as in Their boss paid no atten-
tion to my efforts). Many have both uses (e.g. his, her, which, what, some,
any, either, as in Some like it hot vs. Some guys like it hot). Pronominal
constituents may serve the following autonomous, external functions:
S We tried eating out on the little balcony.
Sp There was nothing we could do about it now.
Od Jack shot himself.
Oi Whoever did you give that book?
Op I take it that you are going to resign.
Cs Jane is someone I regard very highly.
402 Pronominals
Co You can call me that again!

A Where did you hide the doll?
In addition they may function autonomously as conjoints and as preposi-
tional complements:
C JT It was a row between me and someone you don't know.
DEP The scheme was proposed by those working in our department.
In the following example, an autonomous pronoun group somebody else is
down-ranked to determinative status by means of the genitive marker 's:
DEP I borrowed somebody else's book.
Determinative use of pronominale is typically found in noun groups:
(1) [Her smiles] lasted just a fraction of a second too long, (noun group)
(2) [No teacher] likes [all his students], (noun groups)
Occasionally pronouns serve as modifiers:
(3) It cannot be [that bad], (adjective group)
(4) [He himself] cannot speak a word of French, (pronoun group)
The determinative function of pronouns exemplified by Her smiles, No
teacher and all his students is of central importance to our understanding of
the functional domain of pronouns, as we shall see in the next section.
11.1.4. The functional domain of pronominale

Superficially, pronominals seem to serve two main communicative functions:
specification (as in e.g. his decision, this mess, every step, what church, etc.)
and representation of "things' (e.g. they, mine, herself, who, whatever,
everybody, none in examples like They ignored mine, She hated herself, etc.).
These two communicative functions are closely related to the determinative
and autonomous uses mentioned in section 11.1.3 above and thus syntact-
ically distinct. However, if we examine the two communicative functions
more closely we find that they are inextricably related.
To see this, it is useful to recall the discussion of the functional domain of
nominals in section 10.1.4. There the functional properties of nominale were
described in terms of the subfunctions determination, modification, quanti-
fication, categorization, etc. For example, in a noun group like the many poor
artists, the definite article the is a determiner, many is a quantifier, poor is a
modifier and artists is a categorizer and the bearer of the properties ex-
pressed by the other subfunctions. The four subfunctions jointly establish the
referent of the expression.
Preliminaries 403
Significantly, English provides ways of referring to entities where one or

more of the potential subfunctions is not employed, because it is either
irrelevant or can be taken for granted. Thus, obviously, we can often leave
out modification (e.g. the many artists without the adjective poor). Less
obviously perhaps, we can leave out the type of constituent typically serving
as a categorizer and focus primarily on modification: this is what we see in
the nominal use of adjectives (e.g. the poor = 'poor people', the inevitable =
'that which is inevitable'). The modifiers poor and inevitable are here used as
generic categorizers. To achieve more specific reference in expressions with
focus on modification we can use ONE as a substitute for the head noun (as
long as the referent is countable): the poor one/ones. The interpretation of
one or ones is here dependent on e.g. the preceding linguistic context. We
can also focus on quantification alone, as in e.g. Many seem to cope with
things after all. Here, too, the expression relies heavily on the context for the
correct interpretation of the referent.
Last, but not least important, we can focus primarily on determination, as
in e.g. They seem to cope with things after all. The pronoun They may here
represent a full noun group like The many poor artists. The difference
between the two expressions is that while the full noun group establishes the
referent by means of explicit categorization, modification, quantification and
determination, the pronoun simply does so by means of determination and
leaves categorization, modification and, to some extent, quantification
implicit. They can be said to represent e.g. The many poor artists, but of the
subfunctions involved in this noun group, They simply expresses defmiteness
(and general quantification by being a plural pronoun). It is important to note
that the definite meaning of They is not a representationally derived value,
but a value inherent in the pronoun. This becomes clear when we consider
examples like the following:
(1) An old woman and a young man entered the building just before noon. They
were never seen again.
Here They is definite despite the fact that the antecedent An old woman and a
young man is indefinite.
Similarly, the difference between expressions like Look at this mess! and
Look at this! concerns the explicit categorization of the referent. In both
cases we have definite determination.
The picture that emerges is that the speaker may rely on general context or
on information already expressed explicitly and accordingly leave out
subfunctions in a referring expression. In the case of pronouns, they are used
with determinative function not only when they serve as actual determiners
404 Pronominals
(as in Look at this mess!) but also when they serve a representational purpose
more directly (as in Look at thisf),
It is not surprising, therefore, that the notions of reference and referent (cf.
section 10.3.4) are relevant in any discussion of pronouns. Consider e.g.:
(2) I didn't know half the people who were there.
(3) Not having a key to our new home in Mount Street, I had to knock at the
door. Fiona opened it.
(4) Before anyone could stop her, Zelda yelled out.
(5) 'Look at this]' [uttered by someone pointing at the mess in the kitchen]
(2) and (3) are examples of direct non-repetitive anaphoric textual reference:
the antecedent of who is people and the antecedent of if is door. While there
is intrasentential reference in (2) there is extrasentential reference in (3). In
(4), her allows of an interpretation in terms of non-repetitive cataphoric
textual (more specifically intrasentential) reference to Zelda. Finally, in (5)
we have non-textual deictic reference to a specific referent (the mess).
11.2. Central pronouns

11.2.1. Personal pronouns
The table below offers an overview of personal pronouns and the categories
which apply to them:
NUMBER PERSON GENDER CASE
subjective objective
1 / jne
2 - you you
singular
masculine he him.
3 feminine she her
neuter it it
1 - we us
plural 2 - you you
3 - they them
Central pronouns 405
Four categories are relevant to the discussion of personal pronouns: number,

person, gender and case. The first three of these also apply to the derivatives
of personal pronouns: possessive and reflexive pronouns.
A) Number. As with nominals, the pronominal number category comprises
the members singular ana plural. But while the number category is largely
regular in connection with nouns, plural nouns being morphologically
derived from singular nouns (e.g. car/cars), singular and plural personal
pronouns are formally unrelated (except, of course, for the second-person
pronoun you, which is both singular and plural). There is also an important
semantic difference between the nominal and pronominal number category:
while e.g. the plural noun cars is the plural of the singular noun car in the
sense that it simply denotes a plurality of entities denoted by the singular
form, the plural central pronouns do not necessarily denote a plurality of
entities denoted by the singular pronouns. Thus, for example, we is not in
any obvious sense the plural of / (the pronoun denoting the speaker of the
utterance): the plural pronoun does not normally denote a plurality of
'speakers' of the utterance (though in principle it may, e.g. in chants or
petitions involving several individuals). Rather, we typically includes the
speaker plus others associated with the speaker (potentially including the
hearer(s)). In passing we may also note that in formal language we is
sometimes used authoritatively about a singular speaker (e.g. the 'royal we\
as in We are not amused for Ί am not amused', spoken by e.g. the Queen; or
the 'editorial we', as in We therefore propose ... for Ί therefore propose ...',
used by e.g. an author or public speaker).
Singular you represents the hearer: here an ordinary plural interpretation is
more normal. Plural you may represent a plurality of hearers. But it may also
represent a single hearer (or a number of hearers) plus others associated with
the hearer(s).
The third-person plural they differs from the third-person singular pro-
nouns in being gender-neutral: they may refer to a plurality of persons or
things, and if it refers to persons it may refer to male groups, female groups
or mixed groups. Increasingly, they and its derivatives are also used as
gender-neutral terms referring anaphorically to a singular concept, as in
Everybody took their children along (see subsection C on gender below).
The pronominal number category affects subject-predicator concord, but
this topic is best dealt with in connection with the person category, to which
we now turn.
B) Person. In conjunction with the number category, the person category has
formal repercussions for subject-predicator concord. In connection with BE
(cf. section 7.6.2), each person is associated with a distinct form in the
406 Pronominals
present tense: first person am, second person are (which is also used with all
three plural pronouns), third person is. In the past tense, the first- and third-
person singular pronouns take was while the others take were. With other
verbs, person is marked only in the present tense, and only in the third person
singular (e.g. takes vs. take). Referentially, the person category is a deictic
category, the first person being defined in terms of the speaker of the
expression (/ referring to the speaker and we referring to the speaker plus
others), the second person in terms of the hearer (either the hearer alone or
the hearer plus others associated with him or her), and the third person in
terms of referents not directly involved in the communicative act. Note the
occasional use of we about the hearer, basically as an expression of
solidarity but often with a humourous, ironic or condescending effect:
(1) Good morning, Alma, how are we today?
(2) What's this? Are we wearing an expensive new shirt this morning?
As we shall see in the paragraph below on the referential properties of
personal pronouns, there are other extensions of the basic person system.
C) Gender. The gender category applies to the third-person singular
personal pronouns only (plus derived possessive and reflexive forms). The
basic system is as follows: masculine he is used for human males, feminine
she for human females, and neuter (or non-personal) it elsewhere; there is no
common gender term.
This basic system has a number of extensions. As noted in section 10.2.2,
pronominal gender terms are not simply a question of objective sex
distinctions but often reflect the speaker's attitude towards the referent: he
and she may thus be used about animals (e.g. domestic animals, as in He
always barked fiercely at strangers), and especially she is used about other
objects of human affection or concern (ships, cars, countries, etc., as in She's
a fine ship). More generally we can say that he and she are used stylistically
as a means of personification (as in e.g. History has revised her verdict).
Conversely, the neuter term it is occasionally used about a baby (typically as
a marker of dissociation, e.g. It kept screaming all night). Note in this
connection the use of it vs. he or she in examples like the following:
(3) Someone opened the door and entered. It/*She was my mother.
(4) Jack was standing at the top of the stairs. He/*It must be the captain.
While it is used in presentations of identity (as in the first example), he and
she (and plural they) are used in sentences providing further information.
Sometimes both expressions are possible with merely a stylistic difference:
(5) We had invited our new neighbour for dinner. He/It was a young man with
sporty looks.
The traditional use of he (and him as well as the derived possessive pronoun
his and reflexive pronoun himself) as an unmarked, neutral common gender
term when the sex of the referent is unknown, irrelevant or meant to include
both sexes, is still not uncommon although, increasingly, there is a tendency
to avoid the sexist bias, compare:
(6) The reader who works his way through this exposition will be rewarded.
(7) Practically everybody in the place had fallen into the habit nowadays of
looking cautiously over his or her shoulder before he or she spoke.
As the second example shows, there is no simple solution to the problem:
compound units like he or she and his or her are not only cumbersome but,
alas, give linear priority to the male term. The third person plural pronoun
they (plus derived forms) is increasingly used as a common gender pro-form
for singular referents, especially when the antecedent is an indefinite
pronoun:
(8) Everybody/Somebody/Nobody cheered when they heard the news.
These indefinite pronouns are grammatically singular (cf. e.g. Everybody
calls, me Jack) but have plural associations, thus inviting representation by a
plural pronoun. That this usage is not always possible or considered entirely
appropriate is shown by the following examples:
(9) ?Somebody just put their head round the door.
(10) If someone thinks he or she hasn't got love-making completely mastered,
then he or she is likely to try harder, isn't he or she!
(11) 'And almost one in two has a homosexual experience of a genital order at
some time in their lives.' 'In his life,' Adam said. 'Even in the face of the
unbelievable we can still try and be grammatical.'
D) Case. In English, case - apart from the genitive - applies only to
pronouns. Within the class of central pronouns the basic case system is as
follows: there are three cases, the subjective - traditionally referred to as the
'nominative' - the objective - traditionally referred to as the 'accusative' -
and the possessive - traditionally referred to as the 'genitive1. Possessive pro-
nouns will be treated separately in section 11.2.2. The two cases recognized
in connection with the class of personal pronouns, the subjective and the
objective, are assigned on a syntactic basis. Generally, the subjective form is
used only when the pronoun functions as the subject of finite predicators, the
objective form elsewhere (e.g. as direct object or as subject of nonfinite
predicators), e.g.:
(12) She had chosen Wednesday for their flight.
(13) I want him to leave now.
(14) Macon hadn't seen her since his son was born.
408 Pronominals
Note, however, that the subjective form is also used in the normal subject
position in examples like He was hard to beat, which may be analysed in
terms of a discontinuous subject clause (he ...to beat 'to beat him') with he as
the direct object of the infinitive to beat. Note also the special use of the
subjective form of pronominal subjects in absolute clauses (cf. section 8.8):
e.g. She moved forward, he remaining behind.
As subject complement, the subjective case does occur but is generally felt
to be (humourously) hypercorrect unless it is the antecedent of a following
relative clause in which the relative pronoun serves as the subject of a finite
predicator, compare:
(15a) "Who's there?' he called out. 'It is /,' she whispered.
(15b) "Who's there?' he called out. 'It is me,' she whispered.
(16) Actually it is she who rings him.
Compound units pose special problems. A pronoun realizing a conjoint in a
subject compound unit is often found in the objective case in very informal,
spoken language, especially if it realizes the first conjoint:
(17) 'Meet us at noon tomorrow and me and my fat friend will take you to see
whatever you want to see.'
Conversely there is a tendency to use / as the last conjoint (especially in you
and T) irrespective of the function of the compound unit - no doubt as a
result of hypercorrection (cf. Greenbaum & Quirk 1990: 112):
(18) Between you and I, there was some cheating.
Sometimes, if subject function is implied, the subjective case is found even
outside subject position, especially after the prepositions except and but:
(19) No one but he/him laughed, (subject position)
(20) No one laughed but/except he/him, (outside subject position)
(21) Actually it is she who rings him, not /. (outside subject position)
After as and than there is also vacillation:
(22) Lena is much richer than I/me.
The objective case is here the unmarked choice, the subjective case being
slightly formal or affected. Alternatively the speaker may add an operator,
thus explicitly making as or than a conjunction rather than a preposition:
(23) Lena is much richer than / am.
This construction may be chosen to avoid ambiguity, compare:
(24) I hate her more than him. (... than I hate him or than he hates her?)
(25) I hate her more than he does. (... than he hates her)
Note finally that when postmodified by an elaborative nominal (cf. section

10.1.3), the objective form us is an informal alternative to the subjective
form we:
(26) 'You know what us field agents are like,11 said.
E) Syntax. As light-weight constituents, personal pronominal subjects are
not usually separated from the predicator by heavy adverbials (see also
section 5.5.8):
(27) Bob a few minutes later left the building.
(28) *He a few minutes later left the building.
(29) He left the building a few minutes later.
While nominale may precede or follow the adverb in a phrasal verb
construction, unstressed pronouns always precede the adverb (cf. I called him
up / *I called up him). There are also restrictions on pronouns in connection
with full inversion (cf. sections 5.3.5-9), cf:
(30) 'This is an outrage,' said John / *said he.
(31) At his side sat a black Alsatian dog / *sat it.
And, as pointed out in section 3.2.6, while nominal indirect objects always
precede nominal direct objects, pronominal indirect objects may occasionally
follow pronominal direct objects in BrE (e.g. I gave it them).
In compound units, the first person is in polite language realized as the last
conjoint and the second person as the first conjoint: he and I, you and me,
you and he, etc. In the third person singular the masculine precedes the
feminine: he or she rather than she or he - the obvious sexual bias of this
ordering is normally present also in nominal compounds: husband and wife /
men and women /boys and girls /Adam and Eve (but Ladies and gentlemen).
If nominals and third person pronominals are conjoined, the principle of end-
weight (within the compound unit) generally determines the order, compare:
(32) He and Rose disappeared / Rose and he disappeared.
(33) She and an old friend of mine were elected.
(34) Jack and she who was on the phone just before had a brief affair last year.
Personal pronouns only take pre-head dependents in expressions like Poor
him, Clever you, etc. and when used as nouns in examples like:
(35) Is it a he or a she? (i.e. 'a male or female')
(36) I didn't sell my virginity for a sable coat. I've just been me. On the other
hand, that me isn't settled.
However, they do take a broad range of post-head dependents (other pro-
nouns, elaborative nominals, preposition groups, relative clauses):
410 Pronominals
(37) You yourselfmust have realized what was going on.

(38) We professors must make a stand.
(39) We in the English Department cannot support further cuts.
(40) He who stayed the longest fell in love with my wife.
But there are many restrictions: * They professors must make a stand / *He
professor must make a stand / *They who fought in the war oppose the
motion.
F) Reference. Generally first and second person singular and plural
pronouns (/, we, you plus derivatives) have non-textual deictic reference to
the participants of a communicative situation and others associated with
them. Thus, as mentioned in our discussion of the person and number
categories above, / refers to the speaker, not to another constituent in the
linguistic context. We refers to the speaker plus others and you refers to the
hearer or to the hearer plus others (strictly speaking, therefore, the term
'pronoun' is especially inappropriate for these pronouns). A possible
exception to first and second pronominale being purely non-textual deictic
referring expressions is found in cases where we and you have anaphoric
reference to compound units with first and/or second person pronouns as
conjoints:
(41) {You and 1} know what's best, don't we.
(42) {You and Jack} will write to me, won't you.
By contrast, third person singular and plural pronouns (he, she, it, they plus
derivatives) generally have textual anaphoric reference, as in:
(43) "You heard what {Fiona} was saying. She was in East Berlin long enough to
develop strong feelings of friendship.'
Occasionally we come across cataphoric textual reference in subordinate
adverbial clauses optionally placed in sentence-initial position:
(44) Before anyone could stop her, {Zelda} yelled out.
(45) Apart from everything else she'd done to me, {Gillian} had put me off sex.
Only rarely do we come across third person pronouns with specific non-
textual reference:
(46) Tm not going with her, if that's what you think.' [uttered by a child pointing
at a nurse]
The number of people or things embraced by the plural pronouns we, you
and they (plus derivatives) vary from expression to expression (We raised
our children in Birmingham /We Americans cherish our freedom of speech).
Sometimes they are used in a very general sense, somewhat abstracted from
specific referents:
(47) In the twentieth century we have come to take too many things for granted.
(48) I remember thinking as we stood in the baggage hall, this is a bit like the rest
of life. Two of us in a great mass of strangers, and various things to do that
you'\e got to get right, like follow signs and collect your luggage; then>>oM
get looked over by the customs, and no-one particularly cares who you are or
what you're doing there so the two of you have to keep one another cheerful.
(49) I have stopped smoking altogether. You never know what cigarettes will do
to you in the long run.
(50) Somewhere beyond Amiens I had a memory of the car-ferry docks at Calais.
First they send you all round the town and then you get processed into a
system with thousands of other people.
(51) Another thing this young chap does. He talks to himself in his room. I've
heard him. They say these creative people can be a bit potty. But he's got
bags of charm.
Despite the generic-like value of the pronouns in these examples of non-
textual reference, we retains its basic speaker-inclusive meaning while they
retains its speaker- and hearer-exclusive meaning ('those responsible or in the
know'). In examples like (48) to (50), you is all-inclusive and not specifically
hearer-oriented. In (49), the speaker uses you to avoid referring directly and
bluntly to him- or herself. A more formal and impersonal alternative to
general you is one:
(52) One never knows what cigarettes will do to one in the long run.
Note finally that he followed by a restrictive relative clause is sometimes
used generically in the sense 'anyone':
(53) He who betrays our country must be punished.
Such expressions are formal and have an old-fashioned ring to them by
comparison with e.g. those who ...
G) The uses of it. In this final paragraph we summarize the uses of it noted
so far in this grammar (see especially sections 3.2.2 and 3.2.5):
It as provisional subject or object:
(54) // is cruel to let her imagine she is suffering a terrible imitation of her mother.
(55) I thought // unwise to tell my daughter about the affair.
It as a non-referential grammatical prop word, especially in expressions
about weather conditions, time and distance:
(56) Is // snowing again?
(57) It was only two o'clock.
412 Pronominals
(58) It's a long way to Charlotte Pass.

- and in more or less fixed idiomatic expressions:
(59) If they hit // off together, so be it.
(60) This is if!
It in expressions of'identity to be established':
(61) Someone was crying. // was my wife.
(62) "Come on, tell me, what is //?'
and thus also in cleft sentences:
(63) It was my wife who was crying.
Λ as a referential pronoun:
(64) 'So what do you think of {my song} ?' I asked.
Ί think that it's riddled with nauseating self-pity.' (textual extrasentential,
anaphoric reference)
With a collective noun as its antecedent it is in competition with they (cf.
section 7.6.4 [A]).
As a referential pronoun, it may have (part of) a preceding sentence or,
more vaguely, the condition(s) expressed by the preceding linguistic context
as its antecedent:
(65) Being in love makes you liable to fall in love. People think it has to do with
sex, that someone is not doing his duty in bed, or her duty in bed, but I think
this is not the case. It has to do with the heart.
Here it is in competition with the demonstrative pronouns this and that.
After as and than, non-referential // is not used as a subject in examples
like:
(66) I shall act as seems best.
(67) It was a book more rewarding than at first sight might appear.
If, however, the clause contains an infinitive lacking its object or comple-
ment, the subject is not omitted:
(68) The translation is as removed from plain prose as it is possible to be.
(69) He got more than it is possible to get today.
11.2.2. Possessive pronouns

Possessive pronouns can be regarded as personal pronouns in the genitive
case. There are two sets of possessive forms, determinative and autonomous,
corresponding to the specifying and the autonomous nominal genitive (cf.
section 10.3.8 [A]):
NUMBER PERSON GENDER FUNCTION
DET AUT
1 my mine
2 - your yours
singular
masculine his his

3 feminine her hers
neuter its (its)
1 - our ours
plural 2 - your yours
3 - their theirs
Here are some examples showing the determinative and autonomous uses of
possessive pronouns:
Determinative use:
(1) He tacked a note on the door of his little yellow house.
(2) Most of the women were fastening their corsets.
Autonomous use:
(3) He learned that there was probably not another stomach like hers on earth.
(4) He'd fucked up his own life, so he stole mine.
Except for mine and Aw, the autonomous items are formed by adding the
suffix -s to the determinative form. This suffix should not be confused with
the apostrophe s ('s) suffix used in connection with nominal genitives such as
John's (thus while its in its colour is a possessive pronoun, it's is the
contracted form of /'/ is). Note in this connection that its is rarely, if ever,
used with autonomous function without the emphasizer OWN to give it
sufficient weight (cf. Greenbaum & Quirk 1990: 117):
(5) The cat knows that this dish is its own.
A) Possessive pronouns vs. 0/konstructions. Though generally the
determinative possessive forms correspond to specifying nominal genitives
(both constructions expressing definiteness, as in e.g. the old man's hat and
his hat), determinative possessive pronouns have a wider distribution,
414 Pronominals
covering much of the ground occupied by the o/-construction as well as the

specifying genitive in noun groups. Thus while we would normally have to
say the roof of the house rather than the house's roof, we can only say its
roof, not the roof of it. Even objective relations and relations involving
abstract nouns (cf. section 10.3.8 [C.a]) invite expression by possessive
pronoun rather than o/-construction - possibly as a result of the principle of
end-weight, cf. the significance of this idea/its significance; the formulation
of the hypothesis / its formulation', the release of the prisoner or the
prisoner's release/his release, etc.
The o/-construction with a personal pronoun in the objective form rather
than a construction with a possessive pronoun is often used in connection
with:
(i) expressions with little or no possessive meaning: Let's stop this discussion
-1 quite frankly don't see the point of it / The kitchen was one big mess and
in the middle of it was this young kid getting supper ready.
(ii) fixed expressions: the long and s hört of it/on the face of it /by the look
of him /for the life of me/he soon got the knack of it, etc.;
(iii) pronoun groups ending with a personal pronoun like all of them /one of
us /both of you /either of them, etc.: the feelings of either of them rather than
either of their feelings, which strictly speaking is ambiguous.
B) Possessive pronouns as post-genitives. Since determinative possessive
pronouns are definite determiners like nominal genitives, indefinite reference
can only be achieved by means of the post-genitive:
(6) I met a friend of his the other day.
Like nominal post-genitives, pronominal post-genitives typically have
partitive meaning (a friend of his = one of his friends). Pronominal post-
genitives behave like nominal ones with respect to determiner usage (cf.
section 10.3.8 [A]). Note in this connection the lack of partitive association
in groups with a demonstrative determiner expressing the speaker's
(approving or disapproving) interest:
(7) Stuart married that boring little goodie-goodie wife of his.
(8) No doubt she used some rather choice language about this hypothesis of hers.
C) Possessive pronouns vs. the definite article. Since both types of
determiner are definite it is not surprising that there is some overlap.
Generally the distinction is clear: the definite article signals definiteness
only, possessive pronouns signal definiteness plus some relation between
two referents involved (for example, his wife is not only definite but also
expresses a relation between the referents of wife and his). In expressions
referring to parts of the body or clothes associated with someone, possessive

pronouns are used rather than the definite article: He put his hand in his
pocket. The definite article is in such cases used:
(i) in detached, clinically objective expressions such as How is the chest
now? and the rather unusual He put the hand in the pocket (i.e. somebody
else's hand in a separate pocket).
(ii) in preposition groups if the 'possessor' is represented as DONE-TO or
SPECIFIED elsewhere in the clause:
(9) I gave [him]DONE-TO a little poke in the face with my head.
(10) [Bob]00^-10 was hit right on the nose by Jack.
(11) [My sister]SPECIFIED has a terrible pain in the chest.
But there are exceptions:
(12) [Oliver]SPECIFIED needed five stitches in his cheek.
D) Possessive pronouns vs. personal pronouns. As shown in section 8.3,
there is sometimes vacillation between possessive pronouns and the objective
form of personal pronouns when followed by a clause:
(13) It happened without him/his realising what was going on.
E) Possessive pronouns emphasized by OWN. Possessive pronouns can be
emphasized by OWN, as in the following examples:
(14) He apologised for going on so much about Ais own life.
This combination of pronoun and OWN can also be used autonomously, even
in post-genitive constructions without partitive association. In such cases,
however, the determinative form of the pronoun is retained:
(15) I realized that the husky voice was my own.
(16) She hasn't even got a car of her own.
11.2.3. Reflexive pronouns

Below we offer an overview of the reflexive derivatives of personal pro-
nouns in English. In the first and second persons the reflexive pronouns are
formed by adding the singular suffix -self or the plural suffix -selves to the
determinative possessive pronoun. In the third person the suffixes are added
to the objective form of the personal pronoun.
In addition to the reflexive pronouns listed in the table, oneself is used as
the counterpart to one in its generic sense, cf. examples like One does not, on
the whole, permit oneself to attend formal ceremonies to which one has not
been invited.
416 Pronominals
NUMBER PERSON GENDER REFLEXIVE
1 - myself
2 - yourself
singular
masculine himself
3 feminine herself
neuter itself
1 - ourselves
plural 2 - yourselves
3 - themselves
There are three distinct uses of reflexive pronouns in English: a) reflexive, b)

emphatic dependent, and c) emphatic autonomous.
A) Reflexive use. When used reflexively, the pronoun assumes one of the
following functions: Od, Oi, Cs, S of nonfmite P, CJT or DEP of a
preposition, as in the following examples, respectively:
(1) Porter just perched himself 'up in the attic window. (Od)
(2) She bought herself'a new Ferrari. (Oi)
(3) Liz and Roger were not quite themselves. (Cs)
(4) Ύου get out now,' I heard myself saying. (S of nonfinite P)
(5) She blamed me and herself for what happened. (CJT in O:cu)
(6) You see, I used to have this joke with myself. (DEP of preposition)
A defining feature of pronouns used reflexively, apart from their functional
potential, is the fact that they are coreferential with the subject of the clause
within which they occur, as in examples (1) to (3), (5) and (6). If the pronoun
occurs in a subordinate nonfinite clause it is coreferential with the subject of
the matrix clause if it is the subject of the subclause, as in example (4). If in a
nonfinite subclause the pronoun assumes one of the other functions, it is
coreferential with the explicit or implied subject of the subclause, compare
the following examples:
(7) I wanted [her to improve herself].
(8) I wanted [to improve myself].
Transitive constructions with reflexive pronouns as object do not allow

passivization (cf. section 7.4.4 [B]): cf. Jack loves only himself/ *Himselfis
loved only by Jack.
Some verbs always require reflexive objects. Such verbs are called re-
flexive verbs: e.g. ABSENT, BESTIR, INGRATIATE, PRIDE.
(9) He obviously tried to ingratiate himself with his superiors.
Other verbs are used reflexively with a distinct meaning: e.g. APPLY,
AVAIL, CONDUCT, EXERT.
(10) I sometimes got better marks than him, but that was when he chose not to
exert himself, (i.e. EXERT used in the sense 'make a great effort1, not "use1 as
in 'to exert one's influence1)
Reflexivization is obligatory in most cases of coreferentiality. There are how-
ever notable exceptions. In examples like:
(11) Jack loves only Jack.
the lack of reflexivization is a deliberate stylistic choice to emphasize the two
participant roles of the person involved (DOER and DONE-TO) as distinct
and independent.
With some verbs, such as BEHAVE, DRESS, SHAVE, WASH, there is a
choice between strictly intransitive use and transitive reflexive use with only
a slight difference of meaning:
(12) She began to dress (herself).
In preposition groups, the objective form of a personal pronoun is occasion-
ally used instead of the corresponding reflexive pronoun, thus commonly in
constructions expressing space or location:
(13) He didn't seem to notice the decoration above him.
(14) She closed the door behind her.
(15) I have no money on me.
In some cases, the reflexive pronoun is used idiomatically while the personal
pronoun is used in a concrete locational sense, compare:
(16) They were beside me. (concrete location)
(17) They were beside themselves with rage, (figurative meaning)
B) Emphatic dependent use. When used as emphatic dependents, reflexive
pronouns add contrastive meaning to a (pro)nominal constituent:
(18) They didn't appreciate the astonishing privilege they were being granted of
having their troubles treated by Miss Florence Nightingale herself.
(19) He was seeing himself at twelve, standing in Milkman's shoes and feeling
what he himself hud felt for his own father.
418 Pronominals
As in these examples, reflexive pronouns with emphatic dependent use

receive primary stress and often appear in post-head position. But, as pointed
out in section 5.6.3, they also take up other positions, positions identified
elsewhere as adverbial positions in the clause:
(20a) Helen would herself tell me the bad news. (central-M)
(20b) Helen would tell me the bad news herself. (T)
If interpreted as adverbiale rather than dependent group constituents, the
reflexive pronouns here look more like separate referring expressions with
anaphoric, coreferential value.
C) Emphatic autonomous use. Finally, reflexive pronouns are used non-
reflexively and autonomously as emphatic alternatives to the personal
pronouns - typically after prepositions (especially than, as, like, except and
but) or when serving as a CJT in a compound unit:
(21) No one knew this better than himself.
(22) She's about the same age as myself.
(23) No one's to blame but yourself.
(24) He's a pro, just like yourself.
(25) Everybody except herselflaugfred at the joke.
(26) That topic was connected with the terminal row between Sally and myself.
The reflexive pronoun is here often used to avoid having to choose between
the subjective and objective form of the personal pronoun (than he / than
him, as I /as me, etc.).
1 1 .3. Pronouns without a person distinction

11.3.1. Demonstrative pronouns
The demonstrative pronouns in English can be presented in terms of the two
categories applying to them, number and deixis:
singular plural
near this these
distant that those

Pronouns without a person distinction 419
To these four central demonstratives we can add the two locative adverbial
demonstratives here and there.
Demonstratives are emphatic in nature. They have both determinative and
autonomous uses:
(1) At that exact moment they were both indoors.
(2) In this way, the name acquired a quasi-official status.
(3) These couldn't be her children.
(4) He counted those who were late.
Constructions like the following are ambiguous with respect to the
distinction between determinative and autonomous use:
(5) He preferred that chair to this.
Either we here have autonomous use of this referring deictically to some
entity in the context or we have determinative use with suppressed repetition.
The determinative status of this can be made clear by adding one:
(6) He preferred that chair to this one.
Demonstratives assume the usual range of external functions:
(7) Thafs normal, isn't it? (S)
(8) Why am I telling you this! (Od)
(9) I'm going to give those who failed me a little surprise. (Oi)
(10) And that's that. (Cs)
(11) Did you really call her thafl (Co)
(12) I want this and a few other things. (CJT)
(13) How did it ever come to this? (DEP of preposition)
Elsewhere, demonstratives normally serve as determiners (as in that exact
moment, this way, etc.). But in connection with adjectives and quantifiers
such as much and many, the singular demonstratives may serve also as
degree adverbs, indicating a precise amount or measure:
(14) I didn't give her that much.
(15) Do we need this many recommendations?
(16) The worm was this long.
More informally they serve as intensifiers without an association of precise
degree:
(17) The party wasn't that bad.
(18) I was that pleasedl
A) The number category. The singular pronouns this and that are used in
connection with singular and uncountable concepts:
420 Pronominals
(19) This is my favourite dish.

(20) That may prove our worst case ever.
(21) We have to use a little bit of that sugar.
(22) Can you imagine, she didn't want to listen to this music.
These and those are used in connection with plural concepts:
(23) They say these creative people can be a bit potty.
(24) I want two of these and three of those.
Only the singular pronouns are used in connection with aggregate nouns like
FAMILY, GOVERNMENT, TEAM, FIRM, etc.: this family, that government,
etc., not * these family, *those government, etc. However, collectives which
are 'internally countable' (see section 10.4.3) allow of both singular and
plural pronouns, depending on the intended meaning: this people / these
people', that crew / those crew; this staff/ these staff,
B) Reference to persons. When used determinatively, all four demon-
stratives are compatible with personal referents: that girl, this woman, those
neighbours, these composers. When used autonomously, the singular forms
are used only with personal referents in expressions serving as introductions
or identification (in competition with it):
(25) This is my wife.
(26) That was my brother on the phone.
Elsewhere emphatic personal pronouns, not demonstrative pronouns, are
used to refer deictically to singular personal referents:
(27a) Tm not going with her.'
(27b) *Tm not going with that."
The plural forms are used in these contexts too: These are my neighbours /
Those were two of my colleagues. But in addition, these is occasionally used
anaphorically in competition with they (as in All her best friends were there
and these never dare criticize her), and those is used freely with personal
reference when restrictively modified (as in Those who want to continue,
please raise your hands / Those in the know will surely keep it a secret).
C) Deixis. Like the definite article, the demonstrative pronouns are markers
of defmiteness. But unlike the definite article, the demonstratives specify the
referent as near or distant in relation to the speaker, especially in cases of
non-textual reference. The basically deictic nature of demonstrative pronouns
is clear in examples of non-textual reference like the following:
(28) I want these gloves, not those.
(29) Non-native language teacher, holding a book in his hand: 'Repeat after me:
This is a book'
Learner at the back: "That is a book.'
Language teacher: Ί said: This is a book.'
Learner: 'Well, let's just say: It's a book.'
This last example is from an authentic classroom situation, where the learner
had a better intuitive understanding of the deictic nature of the demonstrative
pronouns than the teacher. In both examples, the near pronouns this and
these are used about what is near at hand in relation to the speaker, and the
distant pronouns that and those are used about more distant referents.
The deixis of demonstratives operates not only on a spatial dimension but
also on a temporal one:
(30a) This/*That is the News in English read by ...
(30b) That/*This was the News in English read by ...
(31 a) How 's life these days!
(3 Ib) How was life back in those days!
As we see in these examples, this and these are used about present time and
that and those about past time.
D) Related referential properties. When used with textual reference, the
deictic nature of the demonstratives is often subdued. Only very rarely do we
find examples like (32) (cf. Schibsbye 1970: 208-9), where the interpretation
of the demonstratives rests exclusively on the near/distant distinction:
(32) Work and play are both necessary to health: this gives us rest, and that gives
us energy.
Instead expressions like the following are used: the former ... the latter, the
first... the second.
Normally, with the demonstratives the near/distant distinction is modified or
extended to serve the requirements of textual reference. Both the near and the
distant pronouns are used anaphorically in competition with the lighter
pronoun it:
(33) Oliver's {career}, if that isn't too grand a word for it, had made only a single
movement, and that was downwards.
(34) He's a friendly dog called Poulidor, but {he's now got so old that he's gone
stone deaf}. Both Oliver and I find this terribly sad.
The difference between the near and the distant pronouns is in such cases
very subtle. Often, as in (33) and (34), both variants are possible. The effect
of using the distant pronouns is to direct the hearer's attention to something
mentioned in the preceding linguistic context, while the effect of using the
422 Pronominals
near pronouns is to 'update' something mentioned in what is technically the

preceding linguistic context as part of the current linguistic context, with the
immediacy of present relevance. In other words, with the distant pronouns,
the hearer is guided backwards in the text, whereas with the near pronouns,
the hearer is prompted to consider the antecedent as immediately present in
the communicative situation. The distant pronouns imply a distinction
between the here and now of the speaker and the preceding linguistic
context. With the near pronouns this distinction is more blurred because the
antecedent is presented as part of the current linguistic context.
In practice, though often interchangeable, the distant pronouns tend to be
used with specific, clearly delimited antecedents while the near pronouns
tend to be used with more general antecedents, often approaching a
qualitative value like such, compare:
(35) They sent me Linda. She cost {£100}. That was her price.
(36) {Filthy old lecher, seducer of schoolgirls, abandoner of wife and child} ... A
chap can't expect to get much of a hearing with those labels attached.
(37) It was awful. It was a shouting match. I was just trying to be practical, trying
to express something that I thought came out of my love for Oliver, and he
got all jumpy and hostile. These things don't immediately go away, either.
(38) ... I pass my favourite road sign: ROUTE INONDABLE. Such Gallic
economy. In England it would be DANGER ROAD LIABLE TO
FLOODING. Here, just ROUTE INONDABLE. Then carefully through the
village, and into the welcoming arms of wife and child. How she hugs me,
the iridescent bambino, Little Sal. She clings to me like a wet shower-
curtain. Isn't this the life?
In the last example, it is difficult to identify the antecedent with precision: it
covers in a general way all the good or charming things mentioned in the
preceding context and goes even further back in the passage from which the
example is extracted.
Not surprisingly, the near pronouns rather than the distant pronouns are
used cataphorically, often in contrast to an anaphoric distant pronoun:
(39) Anyway, the point is this: {I didn't go to university}.
(40) Poor old Ollie, up to his mucous membrane in a tub of merde, how
inspissated, how uncheerful... No, actually that's not what I think. What I
think is this. {I love Gillian, she loves me}. That's the starting-point...
The distant pronouns can be used in a cataphoric-like way if what follows
them picks up something from the preceding linguistic context:
(41) Even when I got pregnant it didn't seem to concentrate Oliver's thoughts. 1
tried to explain my concerns to him, and he just said, in a rather pained way,
'But I'm happy, Gill, I'm so happy.11 loved him of course, for that, and we
kissed, and he stroked my tummy which was still as flat as a pancake, and
made some silly joke about the tadpole, and everything was fine for the rest
of the evening. That's the thing about Oliver: {he's very good at making
things fine for the rest of the evening}.
E) Uses without a deictic contrast. The deictic value of the demonstratives
is sometimes replaced by an association of familiarity (plus approval or
disapproval):
(42) I wish him everything, that Stuart: health, hearth, happiness and herpes.
(43) Perhaps he's ashamed of this girlfriend of his.
Idiomatically, demonstratives are used with little or no deictic value:
(44) Oh, we talked about this and that.
(45) That's my big girl!
(46) She finally left him, and that's that.
(47) I know the pedagogue is meant to enthuse his charges by an infectious zest
for learning and all that.
The distant pronouns are used without a contrast to the near pronouns when
followed by restrictive modification:
(48) Let us agree upon the following generality: that those who have inflicted
marriage upon themselves assume such rival guises alternately.
(49) Roger was careful not to mention that which everybody had already guessed.
The construction that which is in competition with the more frequent
independent relative pronoun what with a slightly less precise value (see
section 11.3.3 [B.c] below):
(50) There are various problems with what I'm doing.
When followed by restrictive modification, that and those are in competition
with the one and the ones but not personal pronouns:
(51) Sam changes his chair for that / the one / *it in which his uncle had been
sitting.
(52) The issues in 1960 are no longer those / the ones / *them / *they that existed
in 1935.
In the singular, the expression with that is more formal than the one, while in
the plural those is normal in expressions implying 'established category1, the
ones being somewhat colloquial; cf. also the following example, where the
explicit categorization would render the ones very odd:
(53) The world divides into two categories: those who believe that the purpose of
life is love and everything else is merely an etc.; and those unhappy many
who believe primarily in the etc. of life.
In expressions with specific referents, the ones is normal:
424 Pronominals
(54) Americans are very friendly, and the ones I know are nice to me.
Note that when referring to persons, only the one is possible in the singular:
(55) I was the one who used to be so gloomy about things.
That and those, but not this and these, are used determinatively followed by a
restrictively modified nominal head:
(56) I even had time to give an ironic accent to that crappy bit of the service in
which you promise to 'share' your worldly goods with your partner.
(57) Oliver is one of those people who makes more sense in a context.
The difference between the definite article and demonstrative pronoun in
such constructions is that the latter implies not only definiteness but also
'established category'.
11.3.2. Interrogative pronouns

The central interrogative pronouns in English are:
who, whom, whose
which
what
To this list we can add the following:
when, where, why, how
We retain the term 'pronoun' for these items despite the fact that they usually
correspond to adverbial preposition groups (at what time, in what place, for
what reason, in what way). The archaic items whence 'from where' and
whither 'where1 and the old-fashioned whether used of two possibilities
instead of which, will not be considered here.
A) Syntax. Interrogative pronouns, whose primary function is to form wh-
interrogative constructions, normally take up clause-initial position. In main
clauses they trigger subject-operator inversion unless they serve as subjects
(see section 5.3.3):
(1) Who did you meet? (O)
(2) What is this? (Cs)
(3) Where do you live? (A)
(4) Who wants to go? (S)
Interrogative pronouns may take up terminal position in 'double-barrelled
questions' (Who is who? /He does not know which is which (cf. Jespersen
1933: 307)), as well as in echo-questions (You said WHAT?).
There is no partial inversion in interrogative subclauses:

(5) I asked her who you met / what this was / where you live / who wanted to go.
Note that in corresponding yes-no interrogative subclauses, if ana whether
serve as interrogative conjunctions:
(6) I asked her if/whether you met her.
Interrogative pronouns are rarely used as indirect objects, corresponding
adverbial preposition groups being used instead:
(7a) ? Who/whom did you give the book?
(7b) Who/whom did you give the book to? (or more formally: To whom did you
give the book?)
(8a) *Which of them did you order this radio?
(8b) Which of them did you order this radio for?
(But there is some vacillation: e.g. Who/whom did you serve α four-course
dinner? is acceptable to some native speakers.) In cases of role suppression
in constructions like 'GIVE someone a kiss/smack/nudge etc.1, where the
predicator is fused semantically with the direct object, turning a syntactic
ditransitive S P O O pattern into a semantic monotransitive S P O pattern (cf.
section 7.3.4 [I]), interrogatives may serve superficially as 'indirect' objects,
as in Which of them did you give α kiss?
Autonomous interrogative pronouns are occasionally modified by spe-
cifying or intensifying expressions: where in France, who on earth, why the
hell, etc. Intensification is often provided by adding ever to the pronoun:
whoever, whatever, wherever, etc. Which is frequently followed by an of-
construction: Which of them did it? /Which of your friends are coming?
Similarly, who may be followed by among: Who among them knew what was
going on? Rhetorically, who is also followed by an o/^construction, or even
by a relative clause: Who of us would join the Republicans? / Who that has
heard him can doubt his motives?
The following interrogatives can serve not only autonomously but also as
determiners: whose, which and what, as in e.g. Whose car is this? / Which
book do you prefer? / What strategies have you prepared? Exclamatory what
may serve as a predeterminer: What a load of rubbish you are saying! Only
how can serve as a modifier: How expensive is this car?
B) Categories. Three categories are relevant to the description of inter-
rogatives: case (who vs. whom vs. whose), gender (who vs. what) and what
might be termed 'interrogative scope' (who and what vs. which).
a) Case applies to the series who, whom and whose, the two last being
inflectional variants of the first. While whom is an objective form like e.g.
426 Pronominals
him, its distribution is different. It is obligatory only after a preposition in

clause-initial position, a fairly formal construction: To whom did you give the
book? Very colloquially, the subjective form does occur even after a
preposition but only when the group appears as a complete utterance by itself
(To who?) or in clause-final position (Di shacked up with who?). The reason
we are unlikely to find To who did you give the book is that there is a stylistic
clash: fronting of the whole preposition group is formal whilst the use of the
subjective form who is informal.
The objective form is used also as prepositional complement in dis-
continuous adverbials and as object: Whom did you give the book to? / Whom
did you call last night?, but such constructions are felt to be somewhat
cumbersome and formal. The following sentences thus increase in degree of
formality:
(9a) Who did you give the book to?
(9b) Whom did you give the book to?
(9c) To whom did you give the book?
The genitive form whose is used determinatively and autonomously like
possessive pronouns: Whose shoes are these? / Whose are these shoes?
b) Gender. The gender distinction between personal and non-personal is
relevant to the description of all the three central pronouns who, which and
what. While who (and its derivatives whom and whose) can be used only
about personal referents, which is used about both personal and nonpersonal
referents, cf. the following examples which show determinative and
autonomous use, respectively:
(10) Which car/kid had disappeared?
(11) Which of the cars/kids had disappeared?
Determinative what is also used about both personal and non-personal
referents:
(12) What cars/drivers will make it in time?
Used autonomously, what is used about non-personal referents (but see
subsection c below):
(13) What did he say?
(14) What went wrong?
c) Interrogative scope works on two dimensions: quantitative and
qualitative selectivity. Quantitative selectivity affects the choice of pronoun
in this way: if a question is general, not assuming a limited set of possible
answers, who or what is chosen; if a question assumes a limited number of
alternatives, which is chosen (with an implicit or explicit (»/-construction).

Thus, for example, the difference between:
(15) Who is Roger Wilkinson?
(16) Which (of them) is Roger Wilkinson?
is that (16), unlike (15), requires a specific context with a limited number of
potential bearers of the name Roger Wilkinson. Similarly, in:
(17) What books do you like?
(18) Which (of these) books do you like best?
the first question is completely open, while the second assumes discrim-
ination within a limited set of books. One exception to the rule about who
assuming general selectivity is when it is followed by among, resembling
•which followed by of.
(19a) Who among them would think up such a plan?
(19b) * Which among them would think up such a plan?
Sometimes who and what are used in contexts where, superficially, there
seems to be a limited choice:
(20) What would you like to do: go for a walk or make love to me?
(21) Who did you call, John or Roger?
However, even in such cases, who and what retain their association of
general selectivity, the specified alternatives serving more as examples of
possible answers than as an imposed limitation.
The second dimension of interrogative scope, qualitative selectivity,
distinguishes what from who/whom and which. The difference between them
is often that what is used to query 'kind of referent' whereas who/whom and
which are used to query 'identity of referent', compare:
(22a) Who is your best friend? (Roger!)
(22b) What is your best friend? (A university professor)
(23a) Which years are leap years? (1952, 1956, 1960 ...)
(23b) What years are leap years? (e.g. years in which February has 29 days)
(for these last examples, see Schibsbye 1970: 238). In rhetorical questions,
who is however used not so much about identity but about 'kind of referent1:
(24) Who would do such a thing? (i.e. 'What kind of person would do such a
thing?1)
428 Pronominals
11.3.3. Relative pronouns

The central interrogative pronouns serve also as the central relative
pronouns:
who, whom, whose
which
what
To these we can add the conjunction-like that, the 'zero relative' 0, the
relative adjuncts when, where, why and how and the intensive -ever forms:
whoever, whichever, whatever. The archaic forms whence and whither, the
very formal compounds of where and a preposition (whereby, whereto, etc.)
and the rhetorical -soever forms (e.g. whosoever) will not be dealt with here.
A) Syntax. Relative pronouns, unlike interrogative pronouns, character-
istically serve a double purpose: they signal clausal subordination like
subordinating conjunctions and at the same time they take on a clause
function other than SUB in the subclause (e.g. subject or object), referring
anaphorically to a constituent in the matrix:
(1) You remember the case of {the craftsman} who chipped out a priest's hole
for himself on the ship? (S)
(2) {The second story}, which I pass on without comment, touches on more
delicate matters. (O)
Autonomous function only: who, whom, that, 0, whoever (as well as when,
where, why and how), as in:
(3) They arrested Jeremy Soames, who was on his honeymoon.
Determinative function only: whose, as in:
(4) Why did God preserve this species, whose creation did not reflect
particularly well on its creator?
Either autonomous or determinative function: which(ever) and what(ever);
compare:
(5) He used a gun which he had borrowed from a friend.
(6) He used a gun, which fact bothered the rabbis.
Autonomous relatives may serve as S or Od (as in some of the examples
above) or as Cs, Co, A or DEP in a preposition group, but usually not as Oi:
(7) Roger wasn't quite the speaker that he used to be. (Cs)
(8) He's a bit of a jerk, which some of the girls even call him to his face. (Co)
(9) I met him in the gallery where we used to meet. (A)
(10) Eventually came the day 0 we had been longingôr. (DEP in preposition
group)
(11) * Jane called Roger, whom I wanted to give the book. (*Oi)
Relatives always appear in clause-initial position, as in the examples above,
but note that they are sometimes embedded in larger clause-initial
constituents part of which may precede the relative pronoun, as indeed in this
sentence. Note in this connection the possibility of discontinuity (cf.
Vestergaard 1985: 175 and Preisler 1992: 215-6); compare:
(12a) We interviewed a great many applicants, the majority of whom we rejected,
(continuous)
(12b) We interviewed a great many applicants, of whom we rejected the majority.
(discontinuous)
(12c) We interviewed a great many applicants, who we rejected the majority of.
(discontinuous, informal spoken language)
Apart from intensification in the form of the -ever suffix, relatives are
modified only by the indefinite pronouns both, each and all, as in She had
invited her friends, -who all seemed to enjoy the party.
The distinction between restrictive relative clauses (which help establish
the referent of the antecedent) and non-restrictive relative clauses (which
offer additional information about the referent of the antecedent) is important
syntactically as well as to the choice of relative pronoun (see below):
(13a) The soldiers who were brave ran forward, (restrictive)
(13b) The soldiers, who were brave, ran forward, (non-restrictive)
As far as syntax is concerned, restrictive relative clauses form a DEP closely
tied to its antecedent in a group structure, whereas a non-restrictive clause,
which is syntactically optional and always marked orthographically or
intonationally as a separate information unit, is perhaps best analysed as an
adverbial. For further discussion of the uses of restrictive and non-restrictive
relative clauses, see Bache & Jakobsen 1980.
B) Categories. Relative pronouns are number- and person-transparent in the
sense that it is the number and the person of the antecedent that determine
subject-predicator concord in relative clauses with the relative pronoun as
subject:
(14a) The soldiers who were brave ran forward.
(14b) The soldier who was brave ran forward.
(15) / who am ... I You who are... I He who is ..., etc.
a) The gender category accounts for the distinction between \vho(m) and
•which. As in the examples offered so far, who(m) is used with personal or
430 Pronominals
personified antecedents (including e.g. pet animals, as in Spotty, who was our
first dog, followed little Jane everywhere) while which is used with non-
personal or depersonified antecedents (including sometimes small children,
as in She bore four children, one of which died in infancy. With collective
antecedents (such as FAMILY, GOVERNMENT, etc.), who(m) or which is
used in BrE according to the speaker's view of the referent in terms of
individuals or as a unit (see section 7.6.4 [A]). Determinative which with a
personal antecedent is rare, even in formal language: Mr Johnson, which
gentleman was present,... (Schibsbye 1970: 247).
That and 0 are used with both personal and nonpersonal antecedents: The
woman that/0 you met at my party is Jack's sister; The house that/0 they
built has never been insured.
What is used with personal referents only in determinative use in examples
like She called what friends she had left ('she called the friends that she had
left') where the relative is used independently of an antecedent (on the so-
called independent relatives, see the paragraph on reference below).
Though basically non-personal, which is used indirectly about 'type or kind
of person' when serving as subject or object complement in non-restrictive
relative clauses:
(16) They considered him a frightful bore, which he is.
(17) He's a bit of a jerk, which some of the girls even call him to his face.
Note finally that whose is used not only with personal antecedents (as in /
ran into Jack, whose car had just been stolen) but also with non-personal
antecedents (as in This is the palace whose demolition has been ordered by
the King), though some speakers tend in such cases to prefer the more
cumbersome and formal o/-construction (as in This is the palace the
demolition of which has been ordered by the King).
b) The case category accounts for the distinction between who, whom and
whose. The genitive form whose is used determinatively like the possessive
pronouns and therefore never overlaps with who and whom, which are
always autonomous. The objective form whom is obligatory after a
preposition and generally the preferred form as object and as discontinuous
DEP of a preposition, especially in non-restrictive relative clauses; while the
subjective form who is used as subject and informally as object and DEP in a
discontinuous preposition group; cf. the following examples:
(18) The agent to whom/* to who I transferred all the money has disappeared.
(19) Ian, whom/who I trusted with the money, has disappeared.
(20) Ian, whom/who I transferred all the money to, has disappeared.
(21) The agent who/* whom transferred all the money to me has disappeared.
c) Reference. As already indicated, one of the central functions of relative

pronouns is to refer anaphorically to (part of) a group realizing a clause
function in the matrix, typically a (pro)noun group, as in most of the
examples presented so far, or a preposition group, as in the case of the
relative adjuncts: / met her {at a time} when her parents were still alive; I
found the book {in a place} where no one had thought to look. However,
relative pronouns may also have clausal referents (i.e. serve as sentential
relatives, cf. section 8.10 [C]), which with anaphoric value and what with
cataphoric value, as in the following classic example, in which the sentential
relative clauses very clearly have the value of a disjunct:
(22) {She was late}, which was bad, but what was worse, {she didn't apologize}.
As a parenthetical insertion after a coordinating conjunction which very
occasionally has a cataphoric-like value:
(23) X is reduced to virtually nothing, and, which shouldn't bother us unduly at
this point, {Y is left unaffected}.
The relative clause can here be interpreted as a fronted anaphorically
referring expression, cf. X is reduced ..., and Υ is left unaffected, which
shouldn't bother us unduly at this point. When fronted, the referent of the
relative clause is delimited more precisely as the second conjoint clause only.
Sentential which is commonly used as a determiner, especially in
preposition groups:..., which fact led her to state... / ...,for which reason she
decided... /..., in which case we will have to ...
The close relationship between relatives and conjunctions becomes
apparent when we compare sentential relative clauses to clauses with AS as
SUB:conj (see section 8.10 [C]):
(24a) Being 'selected1 was a mixed blessing, as they soon realized.
(24b) Being 'selected' was a mixed blessing, which they soon realized.
A distinction must be drawn between relatives with the normal textual
(anaphoric or cataphoric) reference and the so-called independent relatives
(see section 8.3). Independent relatives have non-textual reference and can be
interpreted as a fusion of a relative and an antecedent. The following items
can be used in this way: what(ever), whoever, whichever, where(ever),
when(ever), why, how; cf.:
(25) I gave her what was left.
(26) Whoever fails to attend the meeting will be fired.
Independent relatives thus approach the value of a demonstrative or an
indefinite pronoun followed by an anaphoric relative (i.e. 'that which',
'anyone who', etc.). What, unlike whoever, is also used determinatively with
432 Pronominals
personal or non-personal referent, often with an association of 'small

quantity1:
(27) I gave her what oranges we found.
(28) She called what friends she had left.
Who(m) was formerly used as an independent relative, especially in literary
language: Who steales my purse steales trash (Jespersen 1933: 354). Today,
independent who(m) is found only in similar idiomatic expressions like Who
delays pays and with a very restricted set of verbs (CHOOSE, LIKE, PLEASE)
as a very formal alternative to whoever: You may marry whom/whoever you
like. The objective form whomever is rare, if not actually non-existent, even
as a prepositional complement: You may give this book to whoever you want.
Which is used as an independent-like relative only in a clause following a
coordinated clause with independent what:
(29) She held up her hand to stop what she thought was clapping at the back, but
which was two or three women trying to quiet their babies. (Schibsbye 1970:
246)
Arguably, which is here not independent but has anaphoric reference to what.
That is not used independently, except in rare expressions like Handsome
is that handsome does.
The relative adjuncts when, where and why can be used independently:
(30) 8 p.m. is (the time/moment) when you leave.
(31) This is (the place) where I was born.
(32) That was (the reason) why he left.
Relative how is always used independently: This is (*the manner) how you
operate the machine.
Relative whoever, whatever, and whichever are distinguished along the
same lines as the corresponding interrogative pronouns, i.e. in terms of
gender and selectivity:
(33) You may choose whoever you like, (personal; general selectivity)
(34) You may choose whatever you like, (non-personal; general selectivity)
(35) You may choose whatever books you like, (non-personal; general selectivity)
(36) You may invite whatever guests you like, (personal, general selectivity)
(37) You may choose whichever (book, partner) you fancy the most, (personal or
non-personal, limited selectivity)
C) The choice of relative pronoun. There are only few hard-and-fast rules
applying to the choice of relative pronoun. But as a rough first approximation
we can offer the following overview:
Restrictive clauses
Function of pronoun Choice of pronoun Examples
whoAvhich/that The man who/that remained

The table which/that remained
Ο or DEP of that/0/which The man that/01 saw
preposition placed The man that/01 glanced at
finally The table which/that/01 saw
The table which/that/01 glanced at
DEP of preceding whom/which The man at whom I glanced

preposition The table at which I glanced
that/0 He is not the lover that/0 he used to be
Non-restrictive clauses
Function of pronoun Choice of pronoun Examples
who/which Peter, who lives in Lancaster,

Paris, which is the capital of France,
Ο or DEP of who(m)/which Peter, who(m) I know well,
preposition placed Paris, which I know well,
finally Peter, who(m) I glanced at,
Paris, which I am fond of,
DEP of preceding whom/which Peter, to whom I sent the contract,

preposition This room, from which the noise came,
that/0 They consider him a frightful bore,
which he is
Let us now look at some of the finer details in connection with the choice of
relative pronoun:
a) After a preposition only whom and which are used. Continuous
relativized preposition groups are fairly formal:
(38) They chose the material according to the purposeyor which it was intended.
(39) She had lied to Charles, to whom her affair came as a nasty surprise.
b) As complement which is used in non-restrictive relative clauses and that
or 0 in restrictive relative clauses, irrespective of the nature of the ante-
434 Pronominals
cedent, as in They consider him a frightful bore, which he is and He is not the
lover that/0 he used to be.
In other functions, there is much more vacillation in the choice of pronoun,
with the gender-neutral and light relatives that and 0 as common alternatives
to \vho(m) and which, above all in restrictive relative clauses. In general,
considerations of weight, rhythm, syntactic complexity and medium play an
important role. Thus that and 0 are commoner a) when immediately
following the head (pro)noun of the antecedent; b) when the relative clause is
syntactically simple; and c) in informal, spoken language. For example, that
is a more likely pronoun than which (or who) in:
(40) You are right, by the way, to see the animals that fled as the nobler species.
By contrast we are not surprised to find which rather than that in:
(41) They had mobile faces very similar to human beings which you could swear
were about to utter speech.
c) As subject who and which are used in non-restrictive relative clauses. In
restrictive relative clauses, that is sometimes chosen instead of who
especially if the antecedent is modified/determined by a superlative or by all,
any, every, no or only.
(42) They were obliged to advertise, and then select the best pair that presented
itself.
(43) Any student that passes this exam will be admitted to the advanced
phonology course.
As subject, that is a common choice instead of which in restrictive relative
clauses. The difference between them is largely one of medium and style, the
light pronoun being frequent in spoken, informal English, which in written,
formal English:
(44) Now, in the version that has come down to you, the raven has a very small
part.
(45) On the list were several books which could only be studied in the
departmental library.
That is also a common alternative to who and which in cleft sentences (cf.
section 8.4).
As a rule of thumb, 0 is not used as subject. Exceptions are a)
constructions with zero for real subject in existential clauses with there as
provisional subject; b) existential there sentences in colloquial speech; and c)
very colloquial cleft sentences, as in the following examples respectively:
(46) We quickly decided to lie about how many of us 0 there were.
(47) There is someone here 0 wants to see you.
(48) It was John 0 did it, not me!

d) As object or DEP in discontinuous preposition groups in non-
restrictive relative clauses, we find which with non-personal antecedents and
whom (or informal who) with personal antecedents. In restrictive relative
clauses that and 0 are frequent alternatives to who(m) andwhich:
(49) Here's the book that/01 got from my sister.
(50) Here's the book that/0 he keeps quoting from.
(51) My sister helped the girl that/0 we just met.
(52) My sister helped the girl that/01 sent the book to.
With personal antecedents, that or 0 is often chosen to avoid the case
problem of who vs. whom, both of which are also often felt to be somewhat
cumbersome and, in the case of whom, formal. Thus the first example is
clearly preferred to the second:
(53) The man that/01 saw said it was all right to visit you this afternoon.
(54) The man who(m) 1 saw said it was all right to visit you this afternoon.
e) As adverbials relative when and where may occur in both restrictive and
non-restrictive clauses, relative dependent why only in restrictive clauses
with REASON as antecedent. How is only used independently. In restrictive
relative clauses, when and why are often replaced by 0 or that, where only
occasionally so:
(55) I am thinking of the time/morning 0/that he had the ass keel-hauled.
(56) The only reason 0/that he didn't come was that he didn't want to be elected.
(57) The place 0/that we used to meet was unknown to my boss.
(58) *The village 0/that I was bom lies in the Blue Mountains.
The relative adjuncts are normally avoided in cleft sentences:
(59) It was in London 0/that/?where I met her.
(60) It was in the afternoon 0/that/?when I met her.
Relativized prepositional groups are often used instead of the proadverbials
when and where:
(61) My secretary noted the exact time at which the first meeting started.
(62) The room in which the negotiations took place was next to my office.
Note in this connection formal expressions with determinative use of which:
(63) I met her at 10 o'clock, at which time it was too late to warn her.
(64) My boss was late,ybr which reason we had to cancel the meeting.
f) A few extra points: Nonpersonal all, everything and anything are virtually
always followed by that (or 0 outside subject function):
436 Pronominals
(65) All that matters is that you get well.

(66) Everything 0/that he said made her very angry.
In subject function, who is normal after personal all, everyone, anyone and
someone:
(67) Anyone who knows anything about wood could have told him that it was a
hopeless idea.
(68) I am looking for someone who can write up the report for us.
With gender-neutral those, the one and the ones, gender may be signalled by
who and which but of course not by that and 0: those which vs. those who.
Demonstrative autonomous that is nearly always followed by which: I
exchanged my hat for that which he found in the closet.
That instead of which is rare in non-restrictive relative clauses:
(69) The ash trees retain their long and melancholy-looking seeds, that are
sometimes called ash-keys. (Schibsbye 1970: 250)
Outside subject function 0 is often used in restrictive relative clauses if the
subject is a personal pronoun (cf. van Ek & Robat 1989: 169):
(70) The soldier you killed was only a boy.
(71) This is the dictionary I bought yesterday.
11.3.4. Indefinite pronouns

The following four series of pronouns constitute the central system of
indefinites in English:
every everyone everybody everything

some someone somebody something
any anyone anybody anything
no no one nobody nothing none
In addition there are each and all, the dual pronouns both, (n)either and
other(s) and the marginal pronoun one(s). Other marginal items such as the
quantifiers many, more, few, little, enough, etc. and the indefinite pro-
adjuncts (e.g. everywhere, sometimes, somehow) will be touched on in
passing. The -one forms should not be confused with basic items followed by
emphatic one with strong individualizing force as in Every One of them
objected to my proposal.
A) The central system. The four basic items every, some, any and no and
their derivatives are distinguished syntactically and semantically in a number
of ways.
a) Syntax. The derived forms function autonomously only (i.e. as group
heads, S, Oi, Od, Cs, Co and prepositional complement); every and no serve
as determiners only; and some and any have both functions:
(1) Everyone laughed at anything he said.
(2) No director has time to consider every script that comes his way.
(3) Some projects did not get any funding at all.
(4) While I got some of the money, my brothers hardly got any.
The more marginal items expressing place (some-, any-, no-, else-, every-
where), time (sometimes) and manner (somehow, anyhow, AmE someway)
typically (but not exclusively) serve as adverbials:
(5) We couldn't find the flute anywhere. (A)
(6) They sometimes let her down. (A)
(7) Jack somehow managed to finish the manuscript. (A)
(8) Nowhere is as well suited for the project as Brighton. (S)
(9) I comforted the sometimes unhappy sister. (DEP)
The form anyway serves as a conjunct (as in / couldn 't find the flute,
anyway).
When used autonomously, indefinite pronouns allow of postmodification
(e.g. adjectives, preposition groups, relative clauses, else, fcwf-constructions)
but not premodification or determination:
(10) He was toying with something dangerous.
(11) Someone in Paris leaked information to some of his agents.
(12) I don't think we should do anything which might upset them.
(13) I'll give the copy to someone else.
(14) The petitioners were too trepid of this court to let anything but the clear
fountain of truth flow from their mouths.
But note that pre-head dependents qualifying the qualitative meaning are
sometimes possible: hardly anything, nearly everyone, virtually everyone,
etc.
Derived indefinites are occasionally used as nouns and thus allow of pre-H
dependents:
(15) She's a mere nobody.
(16) Bond remarked on the cute little nothing she was wearing.
438 Pronominals
The o/-construction is particularly frequent with some, any and none in

partitive expressions:
(17) I gave her some of the money/books.
(18) Have you read any of John Irving's novels!
(19) None of his students applied for the instructorship.
With something and nothing the »/-construction has qualitative meaning
rather than partitive meaning:
(20) There is something of the mad scientist about him.
(21) She had nothing of her brother's charm.
No and any may serve as dependents in comparative or comparative-like
adjective and adverb groups:
(22) He is no longer my supervisor.
(23) Is her behaviour any better now?
(24) In my view this is no different.
Note in this connection expressions with the positive form good (e.g. Is he
any good? / They are no good) and all/any/none in comparative expressions
with the definite article: We didn't feel any the better for it / We felt none the
wiser.
None is used in connection with too + adjective/adverb and in expressions
with other: I was none too proud of what we had done /He was none other
than the king himself.
Something, nothing and anything approach adverb-like status in expres-
sions like He is something/nothing/not anything like his father.
In colloquial AmE some and any may serve as degree adverbials: Then she
cried some, but that did not bother me any (Preisler 1992: 235).
b) Gender. Among the derivatives, the -one and -body forms are distin-
guished from the -thing forms in terms of the gender category (personal vs.
non-personal). The basic forms some and any, as well as the derived form
none, are used autonomously about both personal and non-personal referents:
Some/None of his cars/friends impressed me. All four basic forms are used
determinatively in expressions with either personal or non-personal referents:
e.g. no sugar / no friend, some cars /some people.
c) Case. The -body and -one forms behave like nouns with respect to case:
they appear in the subjective case (as in the examples examined so far) and in
the genitive case:
(25) It is nobody's fault.
The 'group genitive' (see section 10.3.8) is frequent with else:
(26) It is always somebody else's fault.

d) Number. All the derived forms except none are singular in relation to the
predicator (e.g. Everybody is. happy for you) but often enter concord with
plural central pronouns (Everyone gave their consent; cf. section 11.2.1 [C]).
None takes a singular predicator when referring to something non-countable
(/ was looking for some food but there was none left) and a singular or plural
predicator when referring to countable entities (He gave me many sugges-
tions but none was/were good enough), see also section 7.6.3 [D]. Autonom-
ous some and any are singular when referring to non-countable referents (e.g.
/ was looking for some sugar but there wasn't any): with countable referents
they are plural (/ was looking for some students but there weren 't any).
e) General/restricted. There is no well-established category for the
distinction between the -one forms and the -body forms. Often the two are
felt to be interchangeable: Everybody/Everyone laughed when he fell. The
difference between them is subtle: the -body forms often imply a general
context (as in e.g. Nobody loves me), whereas the -one forms imply a
restricted context (e.g. No one [i.e. no one in the English Department]
agreed with me).
f) Universal/partitive. Moving on to the difference between the four series
of central indefinite pronouns, we operate with the following classification:
<
positive every(-)
negative no(-)
central
indefinite
pronouns
X
assertive some(-)
partitive
\ non-assertive any(-)
While the positive/negative distinction of universal pronouns is intuitively

obvious, the assertive/non-assertive distinction of partitive pronouns requires
some explanation. The assertive forms imply the existence of a referent
whereas non-assertive forms do not. Thus, in positive statements, assertive
forms are normal:
(27) I found some/*any magazines in the drawer.
440 Pronominals
(28) There was someone/*anyone in the kitchen.

By contrast, non-assertive forms are normal in questions and negative state-
ments:
(29) I didn't find any magazines in the drawer.
(30) Was there anyone in the kitchen?
Non-assertive forms are frequent also in conditional clauses:
(31) If you find anything in the drawer, please tell me at once.
However, non-assertive forms are sometimes used in positive statements
with the implication that within the bounds specified there is no limit on the
possible referent. In such cases the non-assertive forms typically occur in
modal contexts and/or are restrictively postmodified; semantically they are
close to the universal, generic every(-) items:
(32) Anyone can beat him in chess.
(33) He did absolutely anything he could think of.
(34) He would tell her anything to discourage her.
Assertive forms are used in questions and conditional clauses with the
implication that there is, or is likely to be, a particular referent:
(35) Did somebody mess up your life?
(36) If you see someone make a pass at her, call me at once.
In negative statements, assertive forms are used if outside the scope of
negation or if there is a strong implication that there is a particular referent:
(37) Some of my friends did not like my new wife.
(38) I couldn't find something wrong with the camera.
g) Reference. When used as determiners, indefinite pronouns do not refer by
themselves but obviously affect the referential properties of the head of the
construction; cf. every student vs. no student.
All the derived -one, -body and -thing forms have non-textual reference
though their scope may be textually restricted by a modifier, as in Someone
in the English Department has complained, or by another (intra- or extra-
sentential) constituent, as in When I visited University of Sussex last year.
everyone was very helpful,
With autonomous some, any and none, textual restriction by means of an
o/Oonstruction is particularly frequent:
(39) Some of my students didn't want to attend any of the courses.
(40) None of my friends could accept the proposal.
The (^construction is occasionally fronted:
(41) Of the many callings that comprise the over-world of intelligence, none
requires as much devotion as that of the sisterhood of listeners.
Despite the strong textual restriction such expressions have strictly non-
textual reference. In practice, the pronouns here function in a determiner-like
way (cf. Some of my students = Some students of mine), letting the
prepositional complement serve as a semantic nucleus in the referring
expression as a whole. With plural meaning some and none are occasionally
used for non-textual reference without any form of textual restriction:
(42) None are so deaf as those that will not hear. (Schibsbye 1970: 260)
(43) Some will say that he did it for the money.
Autonomous some, any and none have textual, anaphoric reference in con-
structions like:
(44) I offered them {some peanuts}. He didn't want any but she took some. She
then asked for {some milk}, but there was none left in the fridge.
Having reviewed the four basic series of indefinite pronouns, we now go on
to briefly examine all, each, the dual pronouns and one.
B) All. Like every(-) and no(-), all is a universal pronoun. It is used
autonomously (with or without postmodification) and (pre)determinatively
about both personal and non-personal, singular and plural referents:
(45) I hope all is well with you and your family.
(46) AH of us agreed to join him for dinner.
(47) Why not take all of me?
(48) Did not Adam give the names to all the cattle?
(49) I shall, in all humility, remain silent on the matter.
With non-textual reference, all is used idiomatically with a general meaning
about non-personal referents (i.e. 'conditions'), as in the first example. With
personal or non-personal plural referents, non-textual all is normally replaced
by everyone/everybody or everything unless there is clear textual restriction
on the potential referents, as e.g. in the second example, where all is
followed by an «»/-construction. In such cases, autonomous all is close in
meaning to determinative all, cf. all of the children vs. all the children, all of
the books vs. all the books. With singular referents, all is used with non-
countable meaning: all humility, all milk, all life, etc. All is used determin-
atively with geographical names conceived as collectives in constructions
like All Brighton were down at the beach.
All is also used as a postmodifier:
(50) I am tired of it all.
442 Pronominals
When the group functions as subject, postmodifying all is in the adverbial

central-M position and thus potentially detached from its head (see section
5.6.3)
(51) They (have) all betrayed her.
All serves as an adverb-like modifier of some adjectives, prepositions and
adverbs:
(52) Jack was all upset about the news.
(53) She was all against my proposal.
(54) A happy solution is now all but ruled out.
(55) He knows that all too well.
(56) The job was done all wrong.
Note finally the use of all in an expression like She was all smiles and fixed
idiomatic collocations: above all, after all, all in all, at all.
C) Each is an individualizing universal pronoun used autonomously (if there
is sufficient textual restriction) or determinatively about both personal and
non-personal singular referents:
(57) The two sisters smiled, each handing me a small envelope.
(58) Each of the books was signed by the author.
(59) I gave each car a thorough examination.
(60) He talked to each member of the department.
Unlike all, each is not used autonomously with non-textual reference unless
there is some textual restriction, as in the first two examples above. When
followed by an o/-construction, autonomous each approaches determinative
each in communicative function: cf. each of the books vs. each book.
Like all and every(-), each refers to all members of a group. But there is a
difference: all focuses on the group as a whole while each picks out the
members of the group individually; every(-) shares some of the value of both
all and each, referring to the individual members of the group collectively:
He told all of us / everyone / each of us the truth. Each is thus near-
synonymous with emphatic every One (which has the association of 'without
exception'). He told every One of us the truth. Unlike all and every(-), each is
capable of referring to groups containing two members only (as in the first
example above).
Like all, each is also used as a postmodifier:
(61) We gave them each a big apple.
When the group serves as subject, postmodifying each is in the adverbial

5.6.3):
(62) They each took a big apple / They have each taken a big apple.
Unlike all, postmodifying each may appear also in terminal position:
(63) We gave them a big apple each.
(64) They took a big apple each.
D) The dual pronouns, which include both, (n)either and marginally other,
presuppose the existence of a class consisting of only two members.
a) Both is the dual counterpart to universal all. It is used autonomously (if
there is sufficient textual restriction) or (pre)determinatively about both
personal and non-personal referents:
(65) My parents turned up later in the evening. Both wanted to see Ann about the
funeral.
(66) Both of the books were signed by the author.
(67) Jack broke down on both occasions.
(68) The referee warned both players.
When followed by an o/-construction, autonomous both approaches
(pre)determinative both in communicative function, cf. Both of the boys dis-
appeared/Both (the) boys disappeared.
Like all and each, both is also used as a postmodifier:
(69) We gave them both a big apple.
When the group serves as subject, postmodifying both is in the adverbial
5.6.3):
(70) They (have) both understood what you are driving at.
b) (N)either. Positive either is the dual counterpart to any, while negative
neither is the dual counterpart to no/none. The two pronouns are used
autonomously (if there is sufficient textual restriction) or determinatively
about both personal and non-personal referents:
(71) Ann and Joan both played, but neither made a good impression.
(72) He was quite happy to sign either of the books.
(73) Either player can make the first move.
(74) Neither painting fetched the expected price.
Either normally means One or the other of two group members', as in the
examples above. In connection with nouns like SIDE and END however, it is
444 Pronominals
sometimes used in an inclusive sense One and the other', approaching the
meaning of both or each:
(75) People were waiting at either side of the street.
(76) There was a little candle at either end of the table.
c) Other is a marginal indefinite pronoun indicating a definite or indefinite
specific alternative. It is used as a (post)determiner or premodifier with or
without determiner: other people, his other daughter, any other suggestions,
all the other neighbours, etc. It is fused with the indefinite article in singular
expressions like another car, another solution, etc. Singular another/the
other and plural others are autonomous with textual or non-textual reference
in constructions like Would you like another? /I put one book in the bag and
the other back on the shelf/ Others disapproved of his departure.
In the singular, the other is used as a contrast to one, in the plural others is
used as a contrast to some: I told him it was one or the other / Some like it
hot, others like it cold.
Other is found in the two reciprocal constructions: each other and (the
more formal) one another: They promised each other/one another to keep in
touch.
E) One has a number of distinct functions:
(i) It is used determinatively or autonomously as a kind of emphatic in-
definite article with singular numerical value in contrast to plural some, other
cardinal numbers (two, three...) and quantifiers like several, many, etc.:
(77) The terrorists were going to execute one hostage an hour.
Its affinity with the indefinite article is obvious also in connection with
proper names; compare: / am looking for a/one Sarah Mortimer (cf. also
section 10.2.5). With contrastive stress it combines with the basic indefinite
pronouns constituting an individualizing alternative to the -one compounds:
Everyone objected / Every "one of them objected.
(ii) As noted in section 11.2.1 [F], generic one (as well as its derivatives
one's and oneself) is used as a formal alternative to you:
(78) One never knows what cigarettes will do to one in the long run.
In AmE, this one may be referred anaphorically to by central pronouns:
(79) One can't always trust himself, now can hel
In tags you is a frequent alternative in colloquial AmE:
(80) One cannot trust anyone anymore, can you!
(iii) One without determination refers in (somewhat formal) expressions to

persons; here one approaches the meaning of'someone':
(81) She was never one to help her friends out of trouble.
(82) He was one who would never forgive but who might forget.
(83) He told her about it, in the casual tone of one commenting on the weather.
One(s) is perfectly normal in definite singular or plural expressions about
persons:
(84) John was the one to consult on sentence analysis.
(85) They are the ones who wouldn't let my sister alone last night.
Note in this connection the more or less fixed expressions: the Holy One, the
Evil One, the little/young ones.
(iv) One is used anaphorically as a pro-form for a whole singular (pro)noun
group ('g-replacive one1):
(86) When she asked for {a cigarette}, I gave her one.
(87) Have you got {another film for this camera}? - Yes, there's one in the
drawer.
(v) Singular one and plural ones are used as a pro-form for a singular or
plural count noun rather than a whole group ('n-replacive one'):
(88) There were three armed {visitors}. The tall one with the glasses had an
almost scholarly air as he tapped the microphone in the manner of lecturers
everywhere.
(89) There were countries in the world which didn't welcome {journalists}, and
who thought that white-skinned ones pretending interest in archaeological
sites were obviously British spies.
Note that n-replacive one(s) serves as a group head, allowing both pre- and
postmodification, as well as determination.
N-replacive one(s) is naturally avoided when the antecedent is non-
countable: Old furniture is sometimes more expensive than new/*new one(s).
It is also normally left out:
(i) after autonomous genitives/possessive pronouns and after own: I much
prefer her car to his /my own /my brother's.
(ii) after numerals and quantifiers like many, several, a few, etc.: They
offered me five books of my own choice but I only took two / There were
plenty of coins in the bag but he only gave her a few.
There is vacillation in the following cases:
446 Pronominals
(i) in indefinite plural or definite singular or plural constructions with two

parallel adjectives, one is often left out, especially in literary language, and
especially if the adjectives express a natural contrast: There were many
Dutch politicians at the meeting but very few German (ones ) /He curiously
preferred poor women to rich (ones) /Everyone likes the old manager better
than the new (one).
(ii) after a comparative adjective one is often left out: Two men entered the
saloon. The younger ordered a drink / This is not a good solution but can
you think of a better?
(iii) after a superlative form one is often avoided outside distinctly colloquial
speech: Having examined the many bottles very carefully, he selected the
cheapest.
(iv) after another, other, last and next, and after ordinals, one is optional but
often avoided in formal language: If you miss the first news broadcast you
can always watch the next one /1 liked the first proposal better than the
second.
With some of the examples with the definite article, it is difficult to decide
whether the adjective is used substantially (cf. section 12.4) or is part of an
elliptic construction.
N-replacive one is normally used:
(i) after the indefinite article + adjective in the positive form: Having
examined the many bottles very carefully, she selected a cheap one. In
distinctly literary language one is here sometimes dropped if the construction
expresses a natural contrast.
(ii) to avoid syntactic ambiguity in cases where the modifier is a potential
nominal head: They were clearly human trails, not just the usual animal
ones.
(iii) after only: Having examined the bottles very carefully, she selected the
only one from Barossa Valley.
12. Adjectivals and adverbals
Adjectivals and adverbals are closely related in a number of ways, the most
important of which are: a) both types of constituent typically express
properties, though in relation to different types of concept; b) the category of
comparison applies to them both; and c) adverbs are often 'de-adjectival',
being derived morphologically from adjectives (e.g. blunt —» bluntly). In the
following we deal first with adjectivals, focussing on three major problem
areas: positional ordering (section 12.2), comparison (section 12.3) and the
substantival use of adjectives (section 12.4). Much of what we say about
adjectivals prepares the ground for the discussion of adverbals in the final
section 12.5.
12.1. Preliminary discussion of adjectivals

Adjectivals typically express properties in relation to the things or entities
expressed by (pro)nominals. For example, in a sentence like The movie was
very boring, the property Very boring' expressed by the complement group is
assigned to the subject group The movie. The same property may be
expressed as an integrated part of the subject group, as in e.g. The boring
movie (took three hours). In this chapter we shall examine the ways in which
adjectivals serve the communicative function of assigning properties.
12.1.1. Adjectives and adjective groups

As will be recalled, adjectivals are either single adjectives or groups with an
adjective as head. Examples:
(1) The bitter rivals spent a long time developing very different systems.
(2) French women are famous for their sensuality.
(3) He was too frightened to tell her the truth.
(4) She is a much more beautiful dancer.
(5) They are clearly taller than she is.
In adjective groups the head adjective accepts a pre-head dependent (as in
very different and much more beautiful) or a post-head dependent (as in
famous for their sensuality and taller than she is) or both (as in too
frightened to tell her the truth). We can thus offer the following description
of the structural potential of adjective groups:
pre-H dependents H post-Η dependents

448 Adjectivals and adverbals
Pre-head dependents are predominantly adverbals (adverbs or adverb groups)

modifying the adjective with respect to degree, as in the examples above.
Frequent degree adverbs found in this position are: VERY, MUCH, QUITE,
EXTREMELY, RATHER, HIGHLY, REALLY, AS, SO, TOO, MORE, MOST,
LESS, LEAST. But we also find pre-head nominale expressing quantity: two
years old /five feet tall / several miles long/ a little reluctant.
Post-head dependents take the following forms: finite clauses (certain that
she will be there), /o-infmitive clauses (eager to try it out), //tan-clauses
(taller than she is) and preposition groups (capable of anything /keen on
playing soccer / better than me). In addition we find the adverb enough in
post-head position (clever enough /fast enough). After WORTH we find
post-head nominals and -ing clauses (worth millions/a fortune/my while/
doing again). Post-head preposition groups and clauses usually express com-
plementation rather than modification (cf. section 6.3.3). In some cases the
presence of a post-head dependent is obligatory: fond of her / bent on doing
it / bound to accept / tantamount to a confession (cf. e.g. *He was fond). In
others the dependent is syntactically optional but if there is one, and if it is
realized by a preposition group, the adjective often selects a certain pre-
position to the exclusion of others: keen on, afraid of, similar to, etc.
We often find discontinuous dependents in adjective groups, especially in
connection with comparison:
(6) Sally is more beautiful than Jane.
(7) Sally is as calm as she was yesterday.
(8) Sally is too frightened to tell him the truth.
Here the post-head dependent is, strictly speaking, a post-head dependent of
the adverb (more/as/too) rather than of the adjective. For discussion of such
cases, see section 5.6.5.
The definite article is used as a pre-head dependent in connection with
comparatives in correlative constructions like the following:
(9) The calmer Sally appeared, the angrier Jack became.
(10) The more important it is to her, the more tongue-tied she becomes.
Such constructions express a systematic relation between two changing
properties. Strictly speaking, the definite article is here associated with the
comparative element (-er, more) rather than with the adjective. The same is
true of constructions with the indefinite pronouns all, any or none followed
by the definite article and a comparative adjective or adverb, e.g. His
explanation left me none the wiser /I hated her all the more because she told
her parents about me.
Preliminary discussion ofadjectivals 449
Adjectives also take the definite article as a pre-head dependent when used
'substantivally' to express entities directly rather than to assign properties to
them: the rich / the inevitable / the stronger / the smallest (for discussion of
this use of adjectives, see section 12.4).
12.1.2. Semantics
Semantically, adjectives express a large range of properties, such as
subjective or emotional evaluation (NICE, GOOD, BAD, BEAUTIFUL), size
(BIG, SMALL, HUGE, LITTLE), shape (ROUND, SQUARE, FLAT, OBLONG),
colour (RED, GREEN, YELLOW, GOLDEN, DARK, LIGHT), nationality
(ENGLISH, CHINESE, FRENCH), age (OLD, YOUNG, ANCIENT), material
(WOODEN, SILVER, SILKY), category (REPUBLICAN, CONGRESSIONAL,
POLITICAL, SOLAR), etc. Many adjectives form pairs of opposites
(antonyms, binary relations): GOOD-BAD, TALL-SHORT, HEAVY-LIGHT,
etc. Some pairs consisting of a positive and a negative member are formed
morphologically by means of affixation (cf. section 7.5.6): HAPPY-
UNHAPPY, CAPABLE-INCAPABLE, SATISFIED-DISSATISFIED, etc. A large
number of adjectives do not enter a binary system but are systemically more
complex (e.g. colour terms and nationality terms) or systemically indeterm-
inate (such as ECONOMIC, RELIGIOUS, LINGUISTIC, INDUSTRIAL, etc),
(see further section 12.1.6).
A number of terms have proved useful for the description of the semantics
of adjectives:
(i) Gradable vs. non-gradable. Gradable adjectives (e.g. NICE, SENSIBLE,
BAD, BEAUTIFUL, etc.) denote scalar properties and thus take degree
adverbs like VERY and EXTREMELY and allow of comparison (e.g. very
beautiful, nicer). Non-gradable adjectives (e.g. ATOMIC, LINGUISTIC,
MEDICAL, OWN, OTHER) denote categorial or determinative properties and
are not normally compatible with intensification or comparison (*very
linguistic, *more atomic, *owner, *very other). Some adjectives are gradable
in one sense and non-gradable in another, cf. a (very) popular politician /
popular culture and a (more) civil young man /civil rights.
(ii) Inherent vs. non-inherent. Inherent adjectives directly ascribe a
property to the referent of the head they modify and may be used equally
well as pre-head modifiers and as subject complements, e.g. a beautiful girl /
the girl is beautiful; a very angry man / the man is very angry. Non-inherent
adjectives relate by way of association to the meaning of the head noun
rather than ascribing a property to the referent as such. Generally, non-
inherent adjectives cannot be used as subject complements. Thus, for
example, an old friend is not necessarily a friend who is old but someone
with whom one has had a friendship of long standing, and a heavy sleeper is
someone who sleeps heavily, not a sleeping person who is heavy. Other
examples of non-inherent adjectives are: a complete idiot / sheer nonsense /
a functional grammarian /foreign policy /young clothes / animate nouns / a
royal hatmaker. Note that non-inherent adjectives sometimes relate to
something rather less than the full head noun (a functional grammarian /a
royal hatmaker}. sometimes to more than the head noun (an old friend —>
Old friendship'), sometimes to something that is only very indirectly
expressed by the head noun (young clothes = 'clothes for young people'). As
can be seen, many adjectives can be used both inherently and non-inherently,
cf. an old horse vs. an old friend and a heavy box vs. a heavy sleeper. In
examples where it is syntactically possible to change a pre-head non-inherent
adjective to subject complement position, it becomes inherent: my old friend
—> my friend is old. For further discussion of this distinction, see Ferris 1993:
19ff.
(Hi) Temporary vs. permanent. This distinction is sometimes used to
describe the difference between examples like the only stars visible and the
only visible stars. In the first construction, visible denotes a temporary
property (i.e. Visible at the moment'), in the second a more permanent
property (i.e. 'normally within sight1). Similarly, a written statement is a
certain kind of statement whereas the statement written is a statement
composed in writing on a particular occasion. In the first instance, written
denotes a standard, enduring property, in the second it has more verbal
character, modifying statement in terms of a situation taking place at a
particular time.
(iv) Stative vs. dynamic. This actional distinction (cf. section 7.2.1)
arguably also applies to adjectives. For example, CAREFUL expresses a
property which presupposes activity while TALL expresses a physical
property independent of activity. Note in this connection that dynamic
adjectives, unlike stative adjectives, may serve as complements in imperative
clauses: Be careful / *Be tall. The fact that both verbs and adjectives can be
described in terms of the distinction between stative and dynamic has led
some linguists to claim that verbs and adjectives belong to the same category
of constituents.
(v) Restrictive vs. nonrestrictive. Dependent adjectives denoting properties
which help the listener establish the referent of the head are called restrictive
adjectives. Whether an adjective is restrictive or nonrestrictive usually
depends on the context. Thus, for example, brave is restrictive in The brave
soldiers ran forward if it singles out a subclass of soldiers in the context (i.e.
if only some of the soldiers, those who were brave, ran forward). If it
describes a property of all the soldiers present in that particular context, it is
nonrestrictive (i.e. 'the soldiers, who were all brave, ran forward'). For further
discussion, see section 10.1.4.
At a later point (section 12.1.6), we examine the classification of adjectives
into descriptive, specifying and classifying adjectives.
12.1.3. Morphology
Adjectives can be divided into several fairly distinct morphological classes.
There are, first of all, a number of simple lexical stems: GOOD, BAD, NICE,
BIG, LONG, OLD, YOUNG, FAIR, CLEVER, NOBLE, MELLOW, etc. But
many adjectives are morphologically more complex in that they are derived
from other word classes, or from other adjectives, by means of affixation.
The two major types are:
(i) denominal adjectives (e.g. BEAUTIELiL RESTLESS, FRIENDLY, BOOR-
ISH, POETIC, HISTORICAL, HUMOROUS, CONSTITUTIONAL)
(ii) deverbal adjectives (e.g. CHARMING., DERIVED, DRUNKEN., REMARK-
ABLE, RESISTIBLE, RESTRICTIVE, DOMINATE).
In some cases, there is more than one derivation from a noun or verb
depending on the intended meaning: FRIENDLY / FRIENDLESS, WORRIED /
WORRYING, etc. In the case of denominal adjectives ending in -ic or -ical,
there is often little or no perceptible difference of meaning, e.g.
POETIC/POETICAL. There is a tendency for the adjective ending in -ic to
denote 'category' and for the one ending in -ical to be gradable and more
descriptive in meaning. A good example of this is ECONOMIC (Of
economies') vs. ECONOMICAL ('careful in spending money, time, etc.'). An
important exception, where the relation is exactly the opposite, is HISTORIC
('notable or memorable in history') vs. HISTORICAL ('pertaining or
belonging to history').
Many deverbal adjectives are present or past participles. Some of these are
fully adjectivalized (e.g. a worried man / a fascinating event) in the sense
that they behave like other typical adjectives (e.g. may be intensified by
VERY). Others preserve more of their verbal character: a rarely/*very heard
opera\ rapidly/*very falling share prices (for discussion, see Huddleston
1984: 30 Iff and 1988: 11 Iff; see also section 7.4.6 on passives vs. adjectival
non-passives).
Usually adjectival past participles are passive in meaning: the deported
prisoners (i.e. the prisoners were deported), the invited relatives (i.e. the
relatives were invited), etc. Occasionally adjectival past participles have
active meaning if they are derived from an intransitive verb denoting a

change of state, e.g. the escaped prisoners (i.e. the prisoners had escaped),
the departed relatives (i.e. the relatives had departed). Some participles allow
of both a passive and an active reading, depending on the nature of the head,
e.g. a returned letter (i.e. the letter has been returned) and a returned soldier
(i.e. the soldier has returned). Often past participles with passive meaning
appear in post-head position: the relatives invited / the prisoners deported. In
such cases, the past participle retains more of its verbal character and can be
interpreted as a reduced relative construction (cf. the relatives who were
invited / the prisoners who were deported) (see also our mention of the
distinction between permanent and temporary properties in section 12.1.2).
Special mention should be made of denominal adjectives ending in the
Verbal suffix' -ed: WINGED (a winged animal), WALLED (a walled garden),
etc. Such adjectives have little verbal character. Instead they imply a
prepositional with-group: 'an animal with wings', 'a garden with a wall', etc.,
cf. also:
(1) These expressions were addressed to a long-legged, short-bodied, small-
headed, white-haired, hog-eyed, runny sort of genius.
A number of adjectives and adverbs contain the prefix a-: AFLOAT, AFRAID,
ALERT, ALONE, ASLEEP, ABROAD, AWAY, etc. One way of distinguishing
between the α-adjectives and α-adverbs is to see whether they can follow the
copula verb SEEM. If they can, they are adjectives (e.g. the patient seemed
afraid/asleep/*abroad/*away). Conversely, if they can follow intransitive
verbs of motion, they are adverbs (e.g. She went *afraid/*asleep/abroad/
away), cf. Greenbaum & Quirk 1990: 131.
We also come across adjectives derived from other adjectives: UNHAPPY,
ATYPICAL, DISHONEST, INCOMPETENT, TJNDE_E-SEXED, HYPERSENSIT-
IVE, P_Q_SX-COLONIAL, EEE-LINGUISTIC, KINDLX, SMALLISH, etc. In
some of these the prefix locally negates the basic adjective, as in the first
four examples mentioned (see sections 7.5.6 and 12.1.2).
Compounding is also an important factor: MUCH-DEBATED, GOOD-
LOOKING, HAND-MADE, etc. Note in this connection the frequent use of
hyphens to signal adjectival status of nominal compounds and other complex
constituents: α remarkable last-ditch effort / other Soviet-bloc countries / the
latest Clinton-for-President movement / a three-year-old girl / a small, what-
else-can-you-expect nod.
12.1.4. The external relations of adjectivals

Adjectivals typically take on one of the following functions:
DEP The clever girls told their anxious mother nothing.

Cs Jane is exceptionally intelligent.
Co They drove him mad.
Naturally, adjectivals also frequently function as conjoints in compound
units with these three functions, e.g.:
CJT She sent him a long and rather boring letter.
Jack is small and rather cute.
They painted the boat yellow and blue.
Adjectivals are often used as complements in verbless adverbial clauses:
Cs They considered Jane (to be) exceptionally intelligent.
If necessary, I can help her.
However disagreeable their presence, you have to let them in.
Adjectives serving as dependents in (pro)noun groups are called attributive
adjectives while adjectives with subject or object complement function are
called predicative adjectives.
In addition to attributive and predicative uses, adjectivals may assume
adverbial function:
A Unhappy with the result, he decided to resign.
I didn't ask him if he liked jazz or, more important, what wine he
preferred.
Dicky hurried in breathless, wearing his new trenchcoat.
Expressionless he drew his head back in again.
He brought his gun loaded.
I was the last person to see her alive.
Adjectivals in this last category are sometimes referred to as 'independent' or
'free1 complements rather than adverbials. They are borderline cases: on the
one hand they assign a property to the subject or object, like genuine subject
or object complements; on the other they are independent optional units
providing supplementary background information, typically about the state
of the entity involved, like many rather more typical adverbials.
In constructions like:
(1) He was out walking this great big dog of his.
(2) She was rather enjoying the nice warm water.
the italicized adjective seems to modify the following adjective rather than
the head noun of the construction and thus assumes adverb-like, dependent
status in relation to the following adjective (i.e. great big = very big; nice
warm = warm in a nice way), see also section 12.2.5 [B] below.
Predicative adjectives (which normally take on subject complement function

after copula verbs) are sometimes used in connection with verbs that are
normally classified as intransitive rather than as copula:
(3) The moon shone bright.
(4) Hope springs eternal in the young man's breast.
(5) The wind blew strong.
In these examples, the italicized adjectives alternate with the adverbs
brightly, eternally and strongly. The effect of using a complement adjective
instead of an adverbial adverb is to change the status of the verb from
intransitive to copula with a corresponding bleaching of the content.
Alternation between complement adjective (describing the subject) and
adverbial adverb (describing the situation) is especially common in
connection with the verbs HANG, LIE, SIT and STAND:
(6) The clouds hung heavy /heavily.
(7) The guests stood silent/silently.
For further discussion of the delimitation of adverbs from adjectives, see
Schibsbye 1970: 160ff.
Like other constituent types, adjectivals may independently assume
communicative functions such as EXC (e.g. Beautiful! /So true! /How sad!).
As the following example shows, adjectivals may also serve as QUE in
heavily context-dependent elliptical constructions:
(8) I was wondering what sort of a gathering it will be. Big? Small? Very
formal? Dinner suit? Sit-down? What's he planning?
Adjectives are sometimes dislocated (cf. section 4.5 on dislocation):
(9) Tricky, that's what writers are!
Note also that adjectives occasionally appear after as and than:
(10) They regarded him as very competent.
(11) She was more hard-working than clever.
An analysis of such construction in terms of preposition groups with
dependent adjectives is possible. But since the adjectives in such
constructions are very predicative-like (cf. They considered him to be very
competent / She was more hard-working than she was clever), an attractive
alternative is to analyse them as C:adj in verbless clauses on a par with
expressions like If necessary. I can help her.
When used attributively, adjectivals typically serve as pre-head
dependents: a friendly doctor / a competent teacher / a very promising
career. Occasionally, however, they appear in post-head position: things
Preliminary discussion of adjectivals 455
Italian /rivers navigable at this time of year. We therefore draw a distinction

between 'attributive pre-head' and 'attributive post-head' adjectivals.
A further distinction should be drawn between those attributive post-head
adjectivals which are restrictive and those which are non-restrictive (often
referred to as appositional). While restrictive attributive post-head adject-
ivals (e.g. the stars visible) are directly attached to the head they modify, like
most attributive pre-head adjectivals, appositional adjectivals provide
parenthetical elaboration (cf. section 12.1.7):
(12) Problems, political, environmental and moral, were dealt with very casually
by the new government.
(13) In the distance he heard a laugh, musical but malicious.
12.1.5. Parataxis and hypotaxis

As noted in the previous sections, adjectivals may realize conjoints in
compound units. The analysis of strings of adjectives is fairly straightforward
in examples like the following:
(1) In the distance he heard a laugh, musical but malicious.
(2) The letter was long and rather boring.
In these examples, we clearly have linked coordination (cf. section 6.2.3)
requiring an analysis in terms of conjoints and coordinators. Unlinked
coordination in post-head and predicative positions is possible but fairly
marked, signalling a 'lingering' literary style:
(3) In the distance he heard a laugh, loud, shrill, malicious.
(4) The letter was long, pathetic, boring.
Again an analysis in terms of compound units seems appropriate but this
time without coordinator conjunctions. In (3) and (4) an alternative is to
analyse the adjectives separately as dependent adjectives directly in relation
to the head noun laugh. But no matter how we analyse them, the adjectives
are paratactically related, i.e. they are at the same level of analysis.
Whenever there are two or more adjectives in attributive post-head or
predicative position, they are always paratactically related.
Parataxis is found also in attributive pre-head position:
(5) He rather enjoyed her dry and light stage kisses.
(6) What she saw was a big, brutal, sweaty boxer.
But in this position, matters are somewhat more complex. Consider the
following examples:
(7) She was visited by a tall dark handsome stranger.
(8) The envious Republican senators complained.

In these examples there is no overt marker of parataxis (coordinators,
commas). The question therefore arises how precisely to interpret the relation
between the adjectives as well as the relation of the adjectives to the head
noun. One possible approach to this problem is to supply the examples with
overt markers of parataxis and see what happens:
(7') She was visited by a tall, dark and handsome stranger.
(81) The envious, Republican senators complained.
What happens here is that we change the meaning somewhat in (8') but not
perceptibly in (7'). This is evidence of parataxis in (7). What we have in (8)
is hypotaxis, i.e. a relation between elements at different levels of analysis.
Without the commas, envious either offers a (non-restrictive) description of
the Republican senators present in the context or it restricts the Republican
senators: some but not all the Republican senators in the context are envious.
In either case, envious assigns a property to Republican senators as a unit
rather than just to senators. As argued in section 4.1.1, a closeness relation
like that between Republican and senators can be captured by analysing the
collocation as a head form stack. By contrast, when envious and Republican
are separated by a comma, as in (8'), the two adjectives relate individually to
the head noun senators or form a compound unit. The implication of this
instance of parataxis is that the envious senators present in the context
happen to be also Republicans. We conclude that in attributive pre-head
position, unlike the attributive post-head and predicative positions, both
hypotaxis and parataxis are possible relations between adjectives.
Note finally that distributive sequences of adjectives, i.e. adjectives
expressing properties in relation to different referents of the same head noun
are always paratactically related and overtly linked: e.g. French and Italian
supporters /professional or non-professional advice.
12.1.6. Descriptive, classifying and specifying adjectives

As a first approximation to the description of adjectives as a word class, we
subclassify them into central and peripheral members. Central adjectives
share the following characteristics:
(i) They are gradable and therefore allow of comparison and intensification
by means of adverbs like VERY and EXTREMELY: these (very) Junny_ plays /
a(n extremely) proud woman / the cold(er) weather / (very) angry teachers /
the (most) beautiful song.
(ii) They occur freely in both attributive pre-head position (as in the
examples above) and in predicative position: these plays are funnv / the
woman was proud /the weather was cold/some of the teachers were angry.
(iii) They often serve as conjoints in linked coordination (or separated by
comma), expressing different properties of the same referent: his ugly and fat
opponent/a high, tinny echo / those Indians are tough and mean-looking.
(iv) They describe rather than classify or define the referent to which they
assign a property.
(v) They typically though not inevitably enter a binary semantic system, e.g.
BIG-SMALL, TALL-SHORT, GOOD-BAD, BROAD-NARROW, SOFT-HARD,
PRETTY-UGLY, CLEAN-DIRTY, STRONG-WEAK, COLD-HOT, HIGH-LOW,
PLEASANT-UNPLEASANT, SIGNIFICANT-INSIGNIFICANT, etc.
Because of their typical semantics, we shall refer to central adjectives as
descriptive adjectives: by assigning a property to a thing they describe the
thing.
Peripheral adjectives do not conform to (some or all of) the criteria
mentioned above:
(i) They are ungradable and therefore do not allow of comparison and
intensification by means of adverbs like VERY and EXTREMELY: *very utter
madness / êxtremely solar energy / *the onlier solution / *a most medical
dictionary / *a more native speaker.
(ii) They do not occur freely in predicative position: *The bed is wooden /
*This mess is entire / *That nomination was presidential / *The student is
former.
(iii) They do not normally enter overtly linked coordination with descriptive
adjectives (or with each other) for the expression of different properties of
the same referent: *their mere and monetary opponent / *my only and
American friend /* those interesting and primary elections. However, we do
find linked coordination distributively assigning properties to different
referents: Danish and Swedish officials /nuclear and solar energy. Note also
that peripheral adjectives may be separated by comma: central, descriptive
adjectives / semantic, interpretational rules. In such cases the property
expressed by the second adjective is presented as reformulation or further
specification of the property expressed by the first.
(iv) They classify or define rather than describe the referent to which they
assign a property.
(v) A few peripheral adjectives enter binary semantic systems like descript-
ive adjectives: NATIVE-FOREIGN, NATIONAL-INTERNATIONAL. However,
most peripheral adjectives do not enter binary systems but are systemically
complex (like medical dictionary /political science / solar energy) or do not
seem to enter a subsystem at all (sheer ignorance / his usual excuse / our
main reason /the veryperson).
There are two main types of peripheral adjectives: classifying adjectives and
specifying adjectives. Classifying adjectives subcategorize the head they
modify. For example, a medical dictionary is a special kind of dictionary and
solar energy is a special kind of energy. Classifying adjectives thus help
establish precisely what sort of thing is involved in the expression. By
contrast specifying adjectives help single out or quantify the referent of the
construction in relation to some context. For example, in his main reason and
my former colleague, the specifying adjectives main ana former have
determiner-like properties.
It is important to note that the division of adjectives into descriptive,
classifying and specifying adjectives is function-based. This means that it is
often difficult to determine the precise subclass membership of adjectives
examined out of context. In each case we have to consider the functional
relationship between the adjective and the head it modifies. For example, in
an English university, English is a categorizing adjective whereas in a very
English response it is a descriptive adjective. Compare also civil behaviour
vs. a civil court ana popular culture vs. a popular actress.
12.1.7. Positional restrictions

Many descriptive adjectives freely allow of both attributive pre-head and
predicative position, but not attributive post-head position in noun groups,
compare:
(1) the new car / the car is new / "the car new
(2) the happy children / the children are happy / *the children happy
Classifying and specifying adjectives normally appear only in attributive pre-
head position:
(3) the medical dictionary / *the dictionary is medical / *the dictionary medical
(4) his main reason / *his reason is main / *his reason main
However, there is a lot of variation with respect to adjectival positions.
In formal language, attributive adjectives may appear in post-head position
and thereby receive end-focus:
(5) He had no patience with problems hypothetical.
(6) His German had the flavour of years long past.

(7) We have other tasks more urgent.
This constituent order is found in indefinite noun groups if the meaning of
the noun is highly general and the adjective subclassifies it, for in such cases
it is natural to give prominence to the constituent with more specific
meaning:
(8) Her weakness for things Italian is quite ludicrous.
Constructions with so + adjective or as + adjective invite post-head position
in connection with indefinite constructions as an alternative to a position
before the indefinite article (cf. section 10.3.6), cf. e.g. so ill a man/a man
so ill and as clever a boy as Jack / a boy as clever as Jack.
In indefinite pronoun groups, the head is so light and general that an
attributive adjective is obligatorily placed in post-head position:
(9) I am looking for something different.
(10) Is there anything interesting on the front page?
Post-head position motivated by the principle of end-focus is also found in
noun groups, definite as well as indefinite, containing dependent deverbal
adjectives:
(11) Some agent unnamed had reached Schlema and gained access to the reports.
As pointed out in our discussion in section 12.1.3 of examples like the
invited guests vs. the guests invited and the deported prisoners vs. the
prisoners deported, post-head position of participial adjectives emphasizes
their verbal character and invites an interpretation in terms of a reduced
relative clause. The same is true of deverbal adjectives ending in -able or
-ible, such as e.g. the navigable rivers vs. the rivers navigable. It is here
tempting to view the rivers navigable as a short version of the rivers that are
navigable. In support of this analysis it may be argued that the rivers
navigable can be expanded into e.g. the rivers navigable by oil tankers at this
time of year, while the navigable rivers cannot be expanded in this way. It
thus appears that many attributive post-head adjectives have clausal
characteristics (see van Ek & Robat 1984: 75). An additional implication in
connection with attributive post-head adjectives is that the property they
express may have temporary application. This can be illustrated by
comparing problems soluble and stars visible with soluble problems and
visible stars. In the former cases, the adjective describes a temporary
property (e.g. 'right now', 'tonight'), in the latter a more permanent, generally
valid property of the head; for discussion see Bolinger 1967.
Adjectives ending in -able or -ible are particularly frequent in post-head

position in noun groups premodified by a superlative or with a semantically
related word such as ONLY:
(12) They had the greatest difficulty imaginable getting there in time.
(13) We must make the best use possible of this attractive offer.
(14) The only room suitable is the one on the third floor.
Sometimes more distinct meanings come to be associated with the position
of the adjective relative to the head noun, compare:
(15a) The headmaster wrote a letter to the parents concerned.
(15b) The headmaster wrote a letter to the concerned parents.
(16a) He thanked all the members present.
(16b) He thanked all the present members.
We leave it to the reader to sort out the different meanings of these examples.
Adjectives with the prefix a- (such as AFLOAT, AFRAID, ALIKE, ALIVE,
ALONE, ASLEEP, AWAKE, AWARE) tend to occur in predicative or
attributive post-head position only:
(17) The children were asleep.
(*the asleep children)
(18) Parents aware of such problems should seek medical advice.
(*aware parents)
ALERT and ALOOF are exceptions to this rule (e.g. He had always shown
aloof hostility to me). Occasionally we find the other α-adjectives in
attributive pre-head position if they are premodified and/or used in a
descriptive sense: the barely afloat oil tanker/a most alive mind.
We often find coordinated classifying or descriptive adjectives in post-
head position for the distributive expression of different entities by the head:
(19) All sorts of problems, political, environmental and moral, would have to be
solved if this bill is passed.
(20) If I'm not allowed to make a profit on popular books, good or bad, I can't
afford to publish less popular books for the discerning minority.
This constituent order, whereby the adjectives are added as an apposition,
serves the purpose of highlighting the adjectives. Discontinuity, separating
the adjectives even further from the head noun, is not uncommon:
(21) If this bill is passed, all sorts of problems would have to be solved: political,
environmental and moral.
Post-head position of coordinated descriptive adjectives is common in
literary style narration:
(22) He was a big man, square-shouldered and virile.

(23) The leaves, so soft and yellow, gave way to the gentle breeze.
(24) There were many sounds, sinister and unidentified, sounds of movement.
Such constructions too are sometimes regarded as reduced relative clauses
(e.g. The leaves, which were so soft and yellow, gave way to the gentle
breeze).
On the whole there is a tendency to avoid heavy pre-head modifiers in
English noun groups. Thus a slight preference is given to constructions like It
was an essay far more interesting than well-written rather than to
constructions like It was a far more interesting than well-written essay. In
general, adjective groups containing dependent complementation or
postmodifiers are placed in post-head position or in predicative position, or
in discontinuous pre- and post-head position (cf. section 5.6.5), as in the
following examples:
(25) Professors keen to take early retirement should contact me immediately.
(26) Further back in the train there were old women with baskets heavy with
home-made vodka and smoked pork sausage.
(27) He was afraid of his sisters.
(28) It falls into a different category from the rest.
When appositional, such modifiers are occasionally separated from their
heads to observe the principle of end-weight:
(29) The lamb had arrived, pink and succulent and tender enough to be eaten with
a spoon.
In contrast to other Germanic languages like Danish and German, construc-
tions of the following type are hardly ever found:
(30) *A by the judge highly valued painting was reported missing,
(or: *A highly valued by the judge painting was ...)
(31) *This resulted in Ά for Wilson unpleasant experience.
(or: *This resulted in an unpleasant for Wilson experience)
Perhaps the best general characterization of the attributive post-head
adjective is to say (with Ferris 1993: 43ff) that it is a mixture of the
attributive pre-head adjective and the predicative adjective. Like the
attributive pre-head adjective it is a dependent in the noun group. But like the
predicative adjective it assigns a property to a thing already fully established.
At the same time, attributive post-head adjectives share a number of
characteristics with the predicative adjective:
(i) We do not find uncoordinated strings of adjectives in these positions (cf.
Ferris 1993: 53f):
(32) "The only book readable missing is one by Twyford.

(33) *The story was long sad.
(ii) Only in attributive post-head and predicative position do adjectives freely
take postmodifiers and post-head complementation, e.g.:
(34a) The experience was unpleasant for Wilson.
(34b) The only experience unpleasant for Wilson was meeting her again.
(34c) *The only unpleasant for Wilson experience was meeting her again.
(iii) Certain adjectives appear only in attributive post-head or predicative
position: the girl is asleep /the only girl asleep / *the asleep girl.
(iv) Attributive post-head and predicative position are not normally taken up
by specifying and (uncoordinated) classifying adjectives:
(35) the former student / *the only student former / *the student is former
(36) the medical student / *the only student medical / *the student is medical
Note finally expressions like court martial, heir apparent, Secretary
General, devil incarnate, body politic, Poet Laureate, president elect, etc.,
which superficially seem to consist of a noun plus a restrictive attributive
post-head adjective. As pointed out in section 10.2.1, such constructions are
best viewed as compound heads.
12.2. Adjectival modification and positional ordering

12.2.1. The functional domain of adjectivals
As we have seen, adjectivals are typically used to express properties in
relation to things or entities. Precisely how they do this depends on their
syntactic realization. As subject and object complements, they denote an
ATTRIBUTE or a RESULT, e.g. The children were unhappy / She got pretty
mad at me / Tyson knocked Bruno unconscious, cf. section 7.3.4. In such
cases the assignment of a property to a thing or entity is the primary
communicative purpose of the sentence. As dependents, adjectivals serve a
secondary communicative role, as in e.g. The happy children returned to the
kindergarten, which primarily reports on the situation of 'returning1, with
happy merely describing one of the two participants of this situation. But the
function of such adjectives is basically the same: to assign a property.
The general term used to describe the functional domain of dependent
adjectivals is modification - one of the subfunctions in the functional domain
of nominals, cf. section 10.1.4. Let us look once again at the relevant part of
the chart offered there for the description of the functional domain of
nominals:
Adjectival modification and positional ordering 463
determination modification categorization (multi-functional)
As this chart shows, the subftmctions of determination, modification and

categorization are arranged in certain specific syntactic zones. In the
following, we shall examine the way in which attributive adjectives occupy
the modificational zone between determination and categorization.
12.2.2. Modificational zones

In section 12.1.6 we recognized three different kinds of adjective: specifying,
descriptive and classifying. As pointed out, these three kinds of adjective are
not (sub)classes in an ordinary sense: an adjective cannot be identified
unambiguously as one or the other in isolation. In each case the relationship
between the adjective and the head noun in a particular noun group must be
carefully examined and interpreted. Rather than speaking of three subclasses
of adjectives, it is more appropriate to operate with three subfunctions of
modification which adjectives may assume in relation to a noun: specifica-
tion, description or classification. This approach is supported by such data as:
CIVIL: civil reply (description)
civil rights (classification)
BLACK: black cloud (description)
black coffee (classification)
PRIMARY: my primary concern (specification)
this primary election (classification)
WILD: a wild party (description)
Australian wild birds (classification)
DANISH: Danish cheese (classification)
a very Danish approach (description)
SECRET a secret plan (description)
the Polish secret service (classification)
WOODEN: wooden bed (classification)
wooden methods (description)
ONLY: an only child (classification)
the only child (specification)
etc.
In these constructions, one and the same adjective functions in two of the
three different ways depending on how it relates to the head noun.
The three subfunctions of modification (specification, description and

classification) are arranged in different syntactic zones between
determination and categorization in noun groups. In other words, they
impose a certain positional order on attributive pre-head adjectives: when
two or more such premodifying adjectives appear together in a noun group,
specifying adjectives precede descriptive adjectives, which in turn precede
classifying adjectives, as shown in the table below. To emphasize the
positional characteristics of the three subfunctions, we refer to specification
as Mod. I (= 'modificational zone Γ), description as Mod. II and classification
as Mod. III. Each of these zones may accommodate zero, one, or more than
one adjective.
.".. :. ·' — '" ;: · - . ;·; · · - ' '.·:·:·
Determination Modification • Categorization |
Specification DeemptiiHB Classification

(Mod. I) ^'QtiiMy <Mod.jn>
the usual sound English stock
her own handsome naval officer
the same beautiful French actress
the next interesting congressional procedure
certain serious organic diseases
the last mighty German attack
the earliest important Aboriginal carvings
many eager medical students
this particular informal linguistic rule
other horrid psychological tricks
In strings of premodifying adjectives belonging to different Mod. zones, it

appears that those adjectives which are closest in function to determination,
viz. specifying adjectives, are placed closest to the determiner and those
adjectives which are closest in function to categorization, viz. classifying
adjectives, are placed closest to the head of the noun group. This means that
there is no strict separation between determination, modification and
categorization but rather a continuum of values from determination to
categorization: from the left determination fades into modification via
specification and from the right categorization fades into modification via
classification. In the middle we have modification at its purest: description.
The term 'central adjective' (which was replaced by 'descriptive adjective')
thus acquires new functional and syntactic significance: a central adjective
appears in central position in the modificational zone and is functionally pure

(i.e. left untainted by determination and categorization).
Adjective order is thus first and foremost a question of the functional
characteristics of adjectives in relation to the head noun. It follows that the
same adjective may appear in different positions depending on its sub-
function. That this is indeed the case is shown in the following data:
(1 a) Scottish popular ballads (HI + III)
(Ib) popular Scottish ballads (11 +HI)
(2a) the first brilliant chapter (I + II)
(2b) the brilliant first chapter (II + III)
(3a) the antique occasional table (III + III)
(3b) the occasional antique table (I + III)
(4a) this good international turn (II + III)
(4b) this international good turn (III + III)
(5a) English dirty books (III + III)
(5b) dirty English books (II + III)
In each of these constructions, the italicized adjective changes its
modificational subfunction in relation to the noun (e.g. popular, which
subclassifies ballads with respect to genre in (la) and describes it in (Ib) -
we leave it to the reader to work out the interpretations of the other
examples). (For a full presentation of this theory of adjective order, see
Bache 1978.)
12.2.3. Inherent Mod. I, Mod. II and Mod. Ill adjectives

Despite the fact that we have defined specifying, descriptive and classifying
adjectives in functional terms rather than as subclasses of adjectives seen in
isolation, it is convenient to regard adjectives as inherent Mod. I, Mod. II or
Mod. Ill adjectives, according to their typical usage. Thus, for example,
GOOD, DIRTY and BRILLIANT are inherent Mod. II adjectives, ENGLISH,
INTERNATIONAL and MEDICAL are inherent Mod. Ill adjectives, and
SAME, FIRST and OTHER are inherent Mod. I adjectives. The point of
treating adjectives as inherent members of Mod. zones is to be able to offer a
description of some of the mechanisms in English for affecting the subfunc-
tion of an adjective apart from changing its position in a string of adjectives.
One effective way of changing an inherent Mod. II or Mod. Ill adjective to
a Mod. I adjective is to give it contrastive stress ("):
(1) I'm taking about the ' big girl not the' small one.
As the specifying force of the adjectives is made clear by prosodic means, it

is possible to keep the normal position of the unstressed, classifying
adjective rather than the one associated with the derived, specifying function:
(2a) I'm talking about the good yellow chair. (II + III)
(2b) I'm talking about the good 'yellow chair. (II +1)
But in some cases, the order is made to conform with the derived function
(cf. Martin 1968: 37-46):
(2c) I'm talking about the 'yellow good chair. (I + II)
Another way of changing inherent Mod. II adjectives, which are typically
gradable (cf. section 12.1.6), to Mod. I adjectives is to subject them to
comparison:
(3) The smarter kids quickly learned how to avoid grounding.
This too may affect the positional order, compare:
(4) The other great achievement was to beat Celtic. (I + II)
(5) The greatest other achievement was to beat Celtic. (I +1)
Inherent Mod. Ill adjectives are occasionally changed to Mod. II adjectives
by means of adverbs of degree superimposing gradability on the originally
classifying meaning of the adjective:
(6) That was a very English remark.
(7) He gave a. fairly political lecture.
Using these techniques of varying the modificational subfunction of
adjectives, as well as positional order, we can offer examples in which the
same adjective appears in all the three Mod. zones:
(8a) the 'black new car (/ + II)
(8b) a small, thin, very black figure (II + II + II)
(8c) strong, sweet black coffee (II + II + ///;
(9a) the more popular Scottish ballads (/ + III)
(9b) popular Scottish ballads (//+ III)
(9c) Scottish popular ballads (III + III)
12.2.4. Structure in and across Mod. zones

There are often more than one adjective in the same Mod. zone:
(la) the first few primaries (I +1)
(Ib) the greatest subsequent numbers (I +1)
(Ic) the only other solution (I + I)
(2a) a new, strange way (II + II)

(2b) the sweet warm stale air (II + II + II)
(2c) a healthy and virtuous girl (II + II)
(3a) one Republican congressional leader (III + III)
(3b) classical Greek drama (III + III)
(3c) tactical nuclear weapons (III + III)
If one wants to ascertain that the analysis of strings of non-central adjectives
(such as (la) to (Ic) and (3a) to (3c)) is correct, one can always try inserting
an inherent Mod. II adjective (such as e.g. INTERESTING or INFLUENTIAL):
if the original adjectives are to the left of the inherent Mod. II adjective in its
most appropriate position (as in the first few interesting primaries) then they
are Mod. I adjectives; if the original adjectives are to the right of the inherent
Mod. II adjective, then they are Mod. Ill adjectives (as in one influential
Republican congressional leader); and if the inherent Mod. II adjective
squeezes in between the original adjectives we have a Mod. I and a Mod. Ill
adjective (as in e.g. the only interesting Greek drama).
Adjectives in Mod. I, in Mod. Ill and in combinations of Mod. I, Mod. II
and Mod. Ill are hypotactically related, while adjectives in Mod. II are
paratactically related. In Mod. II, many adjectives are separated by comma
and/or conjunction (cf. (2a) and (2c) above). If they are not, it is always
possible to separate them by such means without changing the meaning of
the construction:
(4) the sweet warm stale air
= the sweet, warm, stale air
= the sweet, warm and stale air
Furthermore, in many strings of Mod. II adjectives, the order can be reversed
with little or no semantic change:
(5) its dark soft eyes
= its soft dark eyes
(6) a new, strange way
= a strange, new way
(7) equally mindless and vicious types
= equally vicious and mindless types
By contrast, within Mod. I or Mod. Ill, adjectives are not separated by
comma and/or conjunction except to express alternative specification (in
Mod. I) or alternative classification (in Mod. Ill):
(8) the third and smallest class
* the third smallest class
(9) white, Protestant women

Φ white Protestant women
In the third and smallest class, the class referred to could be specified
precisely by third alone or by smallest alone. The construction thus provides
alternative specification of the third class but, in addition, we get the
information that the third class is also the smallest class. By contrast, in the
unbroken sequence the third smallest class, there is complex, progressive
specification. In white, Protestant women, the head noun women is
subclassified by both white and Protestant. The two ways of classifying
women are viewed as parallel, Protestant being an alternative to white. The
implication is 'if white, then Protestant'. By contrast, in the unbroken
sequence white Protestant women, there is complex, regressive classification:
Protestant classifies women, and white subclassifies Protestant women.
Adjectives from different Mod. zones are not normally separated:
(10) an interesting economic strategy (II + III)
(11) the same beautiful girl (I + II)
(12) the first medical dictionary (I + III)
When separation does occur, it is usually semantically significant, as in the
following examples:
(13a) a second context-sensitive rule (I + III)
(13b) a second, context-sensitive rule
(14a) the helpful local dealers (11 +III)
(14b) the helpful, local dealers
Here (13b) and (14b) with the broken sequences differ in meaning from (13a)
and (14a) with the unbroken sequences. (13a), unlike (14b), implies the
existence of a 'first context-sensitive rule1. In (14a) helpful restrictively or
non-restrictively describes local dealers, while in (14b) the description
provided by helpful and the classification provided by local are viewed as
separate, parallel properties of dealers, with the implication that there is
some notional relationship between them (cf. our discussion of the envious(.)
Republican senators in section 12.1.5).
Separation of adjectives is carried one step further in cases where paren-
thetical adjectival insertions elaborate or rephrase a preceding adjective:
(15) a less central, or peripheral, position
(16) purely abstract, but in some sense objective, entities
(17) a further, and much more complex, question
In such cases there are few limitations on the separation of adjectives.
12.2.5. Zone-internal order

In this section we shall comment on the internal order in each Mod. zone:
A) Mod. I: In this zone we can distinguish four major groups of adjectivals
which prove reasonably order sensitive:
(i) Precise and fuzzy ordinal numbers, like FIRST, SECOND, SEVENTH,
NEXT, FINAL, etc.
(ii) Precise and fuzzy cardinal numbers, like TWO, FOUR, FEW, MANY,
COUNTLESS, NUMEROUS, etc.
(iii) Compared forms like older, smaller, better-known, finest, most beautiful,
etc.
(iv) Others, like ONLY, OWN, SAME, OTHER, SUBSEQUENT, FORMER,
MAJOR, SIMILAR, DIFFERENT, MAIN, CHIEF, GENERAL, SPECIFIC,
PRIMARY, CERTAIN, etc.
A string of two or more Mod. I adjectives provides increasing specification.
There is a tendency for the adjectives in the four groups to appear in the
order in which they have just been presented, i.e. Ordinals before cardinals
before compared forms before others':
(1) the first five primaries
(2) the two major categories
(3) six smaller children
(4) the greatest subsequent importance
It is important to note, however, that some Mod. I adjectives may contract
more closely with the definite article for the expression of definite specific
reference, in which case they precede other Mod. I adjectives, irrespective of
their membership of the four groups presented above:
(5) the same particular phenomena
(6) the other six more positive roles
(cf. the six other more positive roles)
(7) the only two utterances
(cf. the two only utterances)
(8) the most beautiful two young ladies
(cf. the two most beautiful young ladies)
Not surprisingly, therefore, positional order is occasionally dependent on the
presence or absence of the definite article, compare:
(9a) three other Nixon associates
(9b) *other three Nixon associates
(9c) the other three Nixon associates

B) Mod. II: In this paratactic, descriptive zone there are few hard and fast
rules for adjective order. Often, as we have seen (cf. section 12.2.4), the
order seems random and can be reversed with little or no change of meaning
(e.g. a harsh thin light vs. a thin harsh light). There are, however, certain
tendencies or preferences:
(i) Short adjectives tend to precede long adjectives (therefore underived
adjectives typically precede derived adjectives):
(10) a deep quiet sleep
(11) a charming, hard-working child
(12) a slight disdainful smile
(ii) Deverbal adjectives tend to precede denominal adjectives:
(13) undulating hilly slopes
(14) predictable wishful distortions
(15) quivering dusky maidens
(iii) Adjectives denoting size, height and length tend to precede other Mod. II
adjectives:
(16) that big, tough guy
(17) a tall, thin creature
(18) long blank periods
(iv) Adjectives denoting size, length and height tend to appear in just that
order:
(19) big, long things
(20) big, high cheek bones
(21) long, low sheds
(v) Emotionally loaded adjectives like BEAUTIFUL, WONDERFUL, LOVE-
LY, HORRIBLE, DREADFUL, NASTY, etc. tend to precede other Mod. II
adjectives, even those denoting size, length and height:
(22) lovely soft hands
(23) a horrible ghoulish enjoyment
(24) a nasty cold wind
(25) a fine big fellow
(26) a terrible small room
Note in this connection that emotionally loaded adjectives occasionally enter
a close relation to the following adjective and assume an almost adverb-like
status:
(27) a great big dog

(= 'a very big dog1)
(28) a tiny little tumour
(= 'a very little tumour')
(29) an awful long trip
(= 'a very long trip')
It is important, finally, to emphasize the fact that adjective order in Mod. II is
extremely variable and cannot be captured by any strict rules.
C) Mod. Ill: In this zone hypotaxis prevails, each adjective (sub)classifying
the following adjective(s) and the head noun. Like the order in Mod. I, the
order in Mod. Ill is relatively fixed. When variation does occur, it affects the
way in which the head noun is (sub)classified, cf. e.g.:
(30a) classical Greek drama
(30b) Greek classical drama
(31 a) the paramilitary Protestant organizations
(31 b) the Protestant paramilitary organizations
(32a) some therapeutic non-hypnotic technique
(32b) some non-hypnotic therapeutic technique
But usually the order is fixed and can be described in terms of certain well-
defined groups of adjectives:
(i) deverbal adjectives, like LEADING, SLEEPING, INTERNALIZED, RECOG-
NIZED, SUSCEPTIBLE, HYPNOTIZABLE, etc.
(ii) adjectives denoting colour, like GREEN, RED, YELLOW, BLACK, etc.
(iii) adjectives denoting nationality, like ENGLISH, FRENCH, CHINESE, etc.
(iv) (other) denominal adjectives, like INDUSTRIAL, PRESIDENTIAL,
NUCLEAR, WOOLLEN, MEDICAL, CULTURAL, POLITICAL, AUTOMATIC,
FISCAL, etc.
(v) nominale serving as premodifiers, like METAL, SILK, FOREIGN POLICY,
TOURIST, AIRLINE, etc.
The linear order is 'deverbal before colour before nationality before
denominal before nominal':
(33) handwritten green pages
(34) a retired Indian Judge
(35) internalized linguistic representation
(36) Mao's supposed deathbed benediction
(37) white American men and women
(38) her pink woollen Dior

(39) a yellow silk handkerchief
(40) the American political system
(41) the increasing Russian military strength
(42) the Democratic foreign policy establishment
(43) an electronic metal detector
Mod. Ill adjectives which denote 'locality' or 'time' often precede (other)
denominal or nominal modifiers but follow deverbal adjectives and
adjectives denoting colour and nationality:
(44) local economic independence
(45) the giant Memphis grain-export firm
(46) daily physical evaluation form
(47) the annual aquatic contest
(48) a mostly white Atlanta district
(49) a growing American national concern
(50) patched-up nineteenth-century houses
Non-inherent Mod. Ill adjectives (i.e. inherent Mod. I or II adjectives)
always immediately precede the nominal head, forming a compound-like
relationship with it:
(51) South-African wild birds
(52) Australian_/osf bowler Jeff Thomson
(53) the various French secret services
(54) unreconstructed cold warriors
(55) this international good turn
(56) key primary states
(57) the five-times-wed former actress
The same place is occupied by one of the inherent determiners when serving
as a modifier: the classifying genitive (cf. section 10.3.8 [A]):
(58) a muggy London summer's day
(59) a standard tourist's guide
(60) black Indian 's hair
(61) artificial silk women's underwear
Finally it should be noted that LITTLE, OLD and YOUNG, which are inherent
Mod. II adjectives, often occur immediately before the head noun with no
intervening Mod. Ill adjectives and seem to enter a kind of compound-like
relation with the head (e.g. little girl, old man, young people). But in strings
Comparison of adjectives 473
of adjectives from both Mod. zones, they usually follow (other) Mod. II
adjectives and precede Mod. Ill adjectives:
(62) a very attractive little American girl
(63) a handsome young Italian doctor
(64) funny old driven snow
12.3. Comparison of adjectives

12.3.1. The basic system of comparison
In section 5.6.5 we introduced the basic system of comparison in English in
order to describe discontinuities in adjective and adverb groups expressing
comparison. Let us begin this section by recapitulating the essential ter-
minology employed for the description of comparison:
(i) We distinguish three members of the category:
positive tall / beautiful
comparative taller / more beautiful
superlative tallest / most beautiful
(ii) There are two types of formal expression: a) morphological comparison
with the suffixes -er and -est.
tall/taller/tallest
and b) syntactic comparison with more and most
beautiful / more beautiful / most beautiful
(iii) A maximal comparative construction consists of three things:
comparative element -er / -est / more / most
adjective (or adverb) tall / beautiful
comparative basis than I am / than me / of them all
Here are some examples:
(1) The younger man was about twenty, with wavy hair and long sideburns.
(2) I liked public phones, they were more private than private ones.
(3) This was our oldest suitcase.
Using the terminology suggested above we can describe the instances of
comparison in examples (1) to (3) in this way:
- in (1) we have morphological comparison; the comparative form younger
consists of the adjective YOUNG and the comparative element -er; there is no
explicit comparative basis.
- in (2) we have syntactic comparison; the comparative form more private

than private ones consists of the adjective PRIVATE, the comparative
element the adverb more and the comparative basis the preposition group
than private ones.
- in (3) we have morphological comparison; the superlative form oldest
consists of the adjective OLD and the comparative element -est\ there is no
explicit comparative basis.
Note that in these descriptions we use the term 'comparative' in two
different ways: it is used specifically to refer to the second member of the
category (taller/more beautiful) but it is also used more generally as the
adjective corresponding to 'comparison' in the terms comparative element
('the element which marks the comparison formally') and comparative basis
('the standard on the basis of which the comparison is expressed'), which are
used in connection with both comparative and superlative constructions.
Comparison involves the ranking of entities on the basis of the degree to
which they possess some property. The kind of comparison dealt with so far,
comparison assigning a 'higher rank' or 'the highest rank', constitutes the
central formal system of comparison in English (in the sense that we here
have regular cases of morphologically marked comparison). It is important to
realize, however, that there are also ways of expressing comparison assigning
the 'same rank' and a 'lower rank' to an entity.
Assignment of the same rank is expressed by as ... as (or sometimes so ...
as).
(4) All right, we'll give you a typing test. Let's see if you are as good as you
claim.
(5) She wouldn't be so mean as him.
In such constructions the first as (or so) is a degree adverb, the second as a
conjunction (as in example (4)) or a preposition (as in example (5)), cf.
section 5.6.5.
Assignment of a lower rank is expressed by means of the irregularly
compared forms of LITTLE, less and least (cf. section 12.3.3 below), plus the
positive form of the adjective (i.e. on analogy with syntactic comparison
assigning a higher rank by means of more and most):
(6) This was a less agreeable place than the railway station.
(Ί) She is the least pretentious professor I know.
(8) Further away there was a smaller, less ostentatious house.
Note in this last example that though the two comparatives in example (8),
smaller and less ostentatious, are paratactically related and semantically
compatible, the former is comparison assigning a higher rank (albeit of

'smallness') whereas the latter is comparison assigning a lower rank.
In addition to the three ways of indicating rank on a property scale ('higher
rank', 'same rank', 'lower rank'), there is a wealth of degree adverbs used in
connection with adjectives to modify the property expressed in terms of
degree but without explicit comparison being involved:
(9) It wasn't very pleasant for Ruth, but she took it so coolly.
(10) The room was rather dark and extremely noisy.
(11) The carving was too asymmetrical for his taste, and the resemblance not
good enough.
12.3.2. Spelling and pronunciation

Certain orthographical changes (very similar to those described in connection
with verbal inflections in section 9.2) occur in connection with morph-
ological comparison:
(i) adjectives ending in written e require only additional -r and -st in the
comparative and superlative, respectively:
fine finer finest
free freer freest
(ii) single final written consonants (except w) are doubled before -er and -est
when the preceding vowel is stressed and spelled with a single letter, com-
pare:
big bigger biggest
narrow narrower narrowest
neat neater neatest
Especially in BrE we get doubling of final consonant also in:
cruel crueller cruellest
(iii) final written -y is normally changed to -i- when following a consonant
but remains unchanged after a vowel, compare:
dry drier driest
grey greyer greyest
Especially in BrE, SHY, SLY, SPRY and WRY keep the y:
shy shyer shyest
Turning now to the spoken language, we begin by noting that the
pronunciation of the suffixes -er and -est are [a(r)] and [ist]. A number of
phonetic changes occur when these suffixes are added to adjectival stems:
(i) final written -r is always pronounced when the suffixes are added:
dear: [dia] [diara] [dianst]
(ii) [ g] in long, strong ana young becomes [rjg] when the suffixes are
added:
long: [log] [loQga] [loggist]
(iii) final syllabic dark / [\] becomes non-syllabic clear 1 [1] when the suffixes
are added:
ample: [aempf] [ajmpla] [aemplist]
In such words morphological comparison thus does not increase the number
of syllables.
12.3.3. Irregular comparison

There are a few irregularly compared forms:
bad/ill worse worst
good/well better best
much/many more most
In some cases there are competing comparatives and/or superlatives:
far farther farthest
further furthest
late later latest
latter last
little less least
lesser
near nearer nearest
next
old older oldest
elder eldest
As adjectives and adverbs both farther/farthest and further/furthest are used
about 'distance1 (e.g. The village is further/farther than the border), though
the -a- forms are not frequent in BrE. Only the -u- forms are used in the
'additional' sense (e.g. further evidence /without further delay).
The regular forms of LATE (later and latest) are used about 'time' (e.g. a
later bus / his latest novel) whereas the irregular ones (latter and last) are
more specialized: in formal language latter is used in contrast to former
about the second of two entities already mentioned (e.g. Joan and Ruth were
both dedicated to the cause: the former offered to take night shifts at the
local hospital, the latter joined our first aid unit). But it is also used to
denote a 'period towards the end' (e.g. In the latter part of the century / her
latter years). While latest means 'most recent', last is used to describe the
final entity in a sequence (cf. her latest/last novel). But last is also used
deictically in contrast to next (cf. last/next week) and with the meaning 'the
one earlier than the one we are talking about' in contrast to demonstrative this
(cf. This meeting is more boring than the last one).
The distinction between nearest and next is similar to the distinction
between latest and last: the regular form nearest is used in the locational
sense whereas next is used with sequential meaning.
The two comparatives of LITTLE typically differ in that less is used quan-
titatively, lesser qualitatively and in certain idiomatic expressions:
(1) They gave me less money than I needed.
(2) This is one of his lesser works.
(3) To invite him along would certainly be the lesser of two evils.
As we saw in section 12.3.1, less and least are furthermore used in connec-
tion with comparison assigning a lower rank.
The compared o- forms of OLD are used generally in comparative
expressions involving the age of entities (e.g. My car is much older than
yours /my sister is much older than she looks /Roger is my oldest colleague
/ my much older sisters). In predicative position only the o- forms are used.
And only the o- forms can be modified, as in much older and the very oldest.
The e- forms are used in attributive pre-head position (as well as
'substantially', cf. section 12.4 below) as an alternative to the o- forms in
expressions involving family relations (e.g. My elder brother is now her
eldest relative). But as the following example shows, the e- forms are
sometimes also used in connection with looser relations:
(4) Two men - dressed in leather jackets and jeans - had followed me. Both
looked like manual workers, but there was a marked age disparity. The elder
man was about thirty-five.
Note finally the occasional use of elder in connection with proper names, e.g.
the Elder Matlock and Pliny the Elder, and in the expression an elder
statesman.
In addition to the irregularly compared forms presented above, there are in
English a number of items that do not enter the normal, three-member system
but which serve as comparative-like or superlative-like forms. Among such
unpaired forms we find adjectives which inherently express comparison,
such as e.g. OTHER, FIRST, SAME, DIFFERENT, SIMILAR, IDENTICAL:
(5) They don't want anyone to know that they are still doing the same nasty
things they did in the bad old days,
There are also a number of items with comparative-like formal character-
istics: INNER, OUTER, UPPER, NETHER, LATTER and FORMER. Unlike
genuine comparatives, these express contrast rather than rank within a
property scale. To indicate superlative meaning in connection with 'locality1,
the suffix -most is used in a number of derivations from different kinds of
root: INNERMOST, OUTERMOST, UPPERMOST, FOREMOST, HINDMOST,
INMOST, TOPMOST, etc.
12.3.4. The choice between morphological and syntactic comparison

Most adjectives require or permit syntactic comparison with more and most.
But some adjectives require morphological comparison and a not insigni-
ficant number permit morphological comparison. Here are the rules:
(i) Monosyllabic adjectives (e.g. BIG, CLEAN, FAST, GREAT, HARD, HIGH,
LOW, QUICK, SMALL, THICK, YOUNG) normally require morphological
comparison:
kind kinder kindest
fine finer finest
Exceptions: JUST, LIKE, REAL, RIGHT, WORTH, WRONG and adjectives
denoting nationality like FRENCH, DUTCH, SWISS, etc. All these require
syntactic comparison.
(ii) Disyllabic adjectives with the stress on the final syllable often but not
invariably take morphological comparison:
polite politer politest
(more polite) (most polite)
profound profounder profoundest
(more profound) (most profound)
However, the following disyllabic adjectives almost always take syntactic
comparison: ANTIQUE, BIZARRE, CONTENT, DEVOUT, ORNATE, those
ending in -esque (e.g. BURLESQUE, GROTESQUE) and those with the prefix
a- (AFRAID, ASLEEP, ALOOF, etc.).
(iii) Disyllabic adjectives ending in an unstressed vowel or syllabic [}]
(written -er, -ow, -y, -le) often but not invariably take morphological
comparison:
clever cleverer cleverest
(more clever) (most clever)
mellow mellower mellowest

(more mellow) (most mellow)
happy happier happiest
(more happy) (most happy)
simple simpler simplest
(more simple) (most simple)
The same applies to COMMON, CRUEL, HANDSOME, PLEASANT, QUIET
and STUPID:
common commoner commonest
(more common) (most common)
Other disyllabic adjectives normally require syntactic comparison:
urgent more urgent most urgent
careful more careful most careful
anxious more anxious most anxious
(iv) Adjectives in three or more syllables require syntactic comparison:
creative more creative most creative
melancholy more melancholy most melancholy
impressionistic more impressionistic most impressionistic
(v) Participles serving as adjectives such as MARKED, DETACHED, FELT,
PLEASED, PRONOUNCED, SHAKEN, MESMERIZED, BORING, DYING,
INTERESTING, SOOTHING, WORRYING, etc. are always syntactically
compared:
pleased more pleased most pleased
boring more boring most boring
(vi) Derived adjectives consisting of an adjectival stem which is normally
compared morphologically and a negative prefix take morphological com-
parison but also permit syntactic comparison:
unkind unkinder unkindest
(more unkind) (most unkind)
untidy untidier untidiest
(more untidy) (most untidy)
(vii) The syntactic comparative of adjectives which permit both morph-
ological and syntactic comparison is especially frequent in predicative
position and when followed by a iAa«-construction:
(1) She was more happy than I thought.
The syntactic comparative is normally required in constructions comparing

two properties in relation to the same entity:
(2) She was more happy /*happier than worried.
Exceptions to this are HIGH, LONG, THICK and WIDE (and their antonyms),
which are followed by a full fhan-c\ause (cf. Schibsbye 1970: 136):
(3) The wall was thicker than it was high.
Morphological comparison is occasionally used for emotional effect in
connection with adjectives normally requiring syntactic comparison:
(4) She would give herself violently; then yawn at the wrongest moment.
(5) He was shy and awkwarder than ever.
(6) 'Curiouser and curiouserl' cried Alice.
Note finally that when compared adjectives are coordinated, morphologically
compared adjectives usually precede syntactically compared adjectives:
(7) He wanted to head the largest and most successfiil publishing house in
England.
(8) The second option was a longer and more tedious route.
Interestingly this order normally coincides with the order that the corres-
ponding positive forms would have (e.g. typically short before long),
whether or not explicitly coordinated: a large successful publishing house/ a
long tedious route (cf. section 12.2.5 [B]).
For a string of adjectives some of which require, others permit syntactic
comparison, this form of comparison is sometimes extended to the whole
string (cf. e.g. Vestergaard 1985: 209):
(9) A solution that is at once more simple, economical and realistic would be to
return to the earlier practice.
12.3.5. The use of compared forms

The positive form of adjectives are used in an absolute sense (i.e. simply to
express a property in relation to an entity with no inherent association of
comparison). By contrast, the comparative form is used in a comparison to
assign to some entity a higher rank on the property scale defined by the
adjective, and the superlative form is used in a comparison to assign the
highest rank to some entity. The term 'property scale' is here used to
emphasize the fact that comparison applies only to gradable adjectives, i.e.
adjectives with scalar meaning. With both comparative and superlative forms
it is important to notice that what is expressed is ranking on the property
scale defined by the adjective rather than the presence of the property
denoted by the positive form of the adjective. For example, in an expression

like Jack is bigger and older than Joan, the speaker ranks Jack higher than
Joan on the property scales of "bigness* and Oldness1. This, however, does not
necessarily mean that the speaker ascribes the properties of T>ig' and Old' to
Jack. Jack can in fact be quite small and young even if he is bigger and older
than Joan.
The comparative is found in expressions of 'comparison between two'.
More specifically it is used:
(i) to assign a higher rank on the property scale defined by an adjective to
one entity (or set of entities) than to another:
(1) He suspected that Henry was cleverer than David.
(2) The incident amused some of the older members.
The comparative basis is either explicit, as in (1), or implicit, as in (2).
(ii) to indicate that an entity ranks higher on a property scale on one occasion
than on another:
(3) She did not look a day older than when they had first met.
(4) Once darkness had fallen the house became more mysterious and sinister.
(Hi) to assign a higher rank to an entity on one property scale than on
another:
(5) His eccentricities were more repugnant than amusing.
(6) The letter was more mischievous than threatening.
The comparative form in English is not infrequently used in an absolute
sense with an association of '(fairly) high degree' rather than "higher rank'. In
some examples inviting this analysis, the comparison signalled by the
comparative form is left unspecified in the context and therefore implies a
very general comparative basis: e.g. This company produces larger tents (i.e.
'larger than most') and Our dog seems to like older people (Older than
average'). In such examples the distinction between 'fairly high degree' and
'higher rank' is difficult to draw precisely. In other examples we have
comparative forms expressing a more genuine absolute meaning, namely
'contrast', as in the lower classes, an upper lip, the outer walls, etc.
A gradually increasing degree of a property is expressed by repeating the
comparative form (or marker) in a compound unit:
(7) She sounded angrier and angrier.
(8) He became more and more irritating.
Turning now to the superlative we note first of all that this form is found in
expressions of 'comparison between more than two*. More specifically it is
used:
(i) to assign a higher rank on the property scale defined by the adjective to
one entity (or set of entities) than to any other (sets of) entities in a
comparison, thus in effect ranking the entity as 'number one':
(9) This is the oldest publishing firm in the country.
(10) He bought the house on the highest mortgage obtainable.
(11) She was the youngest actress of them all.
(11) to indicate that an entity (or set of entities) ranks higher on the property
scale on one occasion than on any other occasion:
(12) Children are happiest when they know from the start who is boss.
(13) Do you have any political enemies, using that word in the widest sense?
Several comments are pertinent in connection with the uses of the super-
lative. First of all, it is important to draw a distinction between 'comparative
basis' and 'scope of comparison*. In examples (9) and (10), unlike example
(11), the comparative basis is merely implied (because there is no reference
to the other entities involved in the comparison), but both examples contain a
specification of the scope within which the other entities are to be found (in
the country / obtainable). Secondly, examples like (11), where we find an
explicit comparative basis (of them all) in connection with a pre-head
superlative, are not frequent. More often one comes across examples where
the superlative assumes head function (cf. section 12.4 below) with an
explicit or implied o/"-construction as the comparative basis:
(14) The incident brought back to mind unbidden the greatest of all Thames
tragedies.
(15) She was by far the youngest.
In both head and pre-head position the superlative requires a definite
determiner.
The morphological superlative is occasionally used in an absolute sense,
i.e. with an association of 'exceptionally high degree' rather than the usual
association of'highest rank':
(16) Her face expressed the liveliest gratification.
The syntactic collocation of most and an adjective (or adverb) is often used
for such purposes, especially in connection with the indefinite article:
(17) It was a most indecent proposal.
(18) I had a most interesting conversation with her the other night.
The substantival use of adjectives 483
Note in this connection also standard expressions like: With best wishes, at
last, at least, my dearest Sarah, etc.
The superlative is often informally used instead of the comparative in
expressions of 'comparison between two' when there is an explicit or implied
o/-construction as the comparative basis:
(19) Joan is the more/most considerate of the (two) sisters.
(20) He accepted the smaller/smallest sum.
12.4. The substantival use of adjectives

12.4.1. What is meant by 'substantival use1?
Traditionally, adjectives are said to be used substantially when they serve as
heads in entity-expressing constructions like the following:
(1) The new images were frightening. They ranged from the grotesque to the
obscene to the simply horrible.
(2) Pina was a formidable Italian matron who welcomed the rich and famous
while ruthlessly pruning from her clientele those of lesser appeal.
(3) Pauline is mildly piqued at the film's suggestion that it was Astrid rather than
Stu who was the more influential of the pair.
(4) James hadn't told us the -worst yet.
We have retained the term 'substantival use' because it aptly reflects the fact
that when adjectives are used in this fashion, 'substance' is in some (concrete
or abstract) sense added to the normal adjectival expression of a property: a
property becomes an entity. In example (1), the three italicized constructions
express general abstract properties as if they -were entities (e.g. the grotesque
= 'something grotesque' or 'that which is grotesque'). In example (2), the rich
and famous refers to 'rich and famous people' as a general kind, and in
example (3), the more influential refers to a particular person. Finally in
example (4), the worst expresses a specific entity (e.g. 'the worst informa-
tion', 'the worst aspect of the case'). In each of these instances, the adjective
expresses a property (like adjectives used non-substantivally) but the
property is not related to an entity expressed by a separate constituent but is
itself used to establish an entity.
Substantially used adjectives are a mixture of adjectives and nouns. Like
nouns they are used to express entities, and they normally take the definite
article, as we have seen. Like adjectives they can be modified by adverbs (cf.
the simply horrible) and permit comparison (the more influential / the worst).
Substantially used adjectives should be delimited from:
(i) Lexical items that serve as both adjectives and nouns, such as CRIMINAL,
GERMAN, IMBECILE, PRIMARY and SAVAGE. Unlike these, substantially
used adjectives do not normally accept the indefinite article, nor do they
accept the plural -s form or the genitive, e.g. a criminal / criminals / this
criminal's vs. *a more influential/*more influentials / *this influential's.
(ii) Elliptical constructions with adjectives missing their head nouns:
(5) This is not a good solution but can you think of α better (one)?
(6) There were many Dutch politicians at the meeting but very few German
(ones).
(7) Old furniture is sometimes more expensive than new.
In these examples we have anaphoric ellipsis (cf. section 4.2.2). Unlike
substantially used adjectives, better, German and new are here used as
parallels to premodifying adjectives (good/Dutch/old) and invite the use of
N-replacive one if the entities involved are countable, as in examples (5) and
(6), cf. section 11.3.4 [E].
(Hi) Pronoun groups with a demonstrative pronoun as head, cf. the injured
vs. those injured. The former is an example of substantival use of INJURED
expressing a category of people, the latter is a pronoun group with those as
head having specific reference.
12.4.2. Generic and specific reference

Substantially used adjectives may have either generic or specific reference.
Generic reference is typically found in connection with:
(i) Descriptive (i.e. inherent Mod. II) adjectives expressing an abstract entity:
(1) Having put up with Mrs Thatcher for 10 years he clearly felt he could no
longer defend the indefensible.
(2) 'Look,1 she retorted angrily, Ί know I'm good, but even I can't manage the
impossible at such short notice!1
(3) Dale didn't look well for this interview; his waxy pallor had slid over into the
sickly.
(ii) Descriptive (i.e. inherent Mod. II) adjectives expressing a category of
people:
(4) He accepts opponents without distinction - the stupid, the wily, the vain, the
cautious, the desperate, the hopeless.
(5) If the police waste time suspecting the innocent they'll have less chance of
catching the guilty.
(6) They know how to change the way people look. They make the pretty
beautiful, the passable pretty, and the ugly interesting.
The substantival use of adjectives 485
There is normally a clear association of plurality involved in such

expressions, and as subjects they enter plural concord with the predicator
(e.g. The rich tend to prefer this area).
Certain nationality terms and participles typically appearing in Mod. Ill
may also be used substantially to refer to a category of people: the English
/ the French / the Irish / the Welsh /the Dutch / the Swiss / the Chinese / the
Indonese / the Japanese / the Portuguese / the living / the dying / the injured.
Specific singular or plural reference is typically found in connection with:
(i) Comparatives and superlatives (i.e. forms usually appearing in Mod. I):
(7) The older of the two men turned out to be the poetry editor and the younger
the accountant.
(8) She would be the less regarded of the two, the less popular, the poor
relation.
(9) These clubs were by far the most convenient in London.
(10) The incident brought back to memory the greatest of all Thames tragedies.
The use is particularly frequent in connection with an explicit or implied of-
construction as the comparative basis.
(11) Certain other adjectives typically appearing as Mod. I adjectives:
(11) She promised to give me a call some time, but I knew that it would never be
quite the same.
(12) I went up to him and asked for the usual.
(13) In the following we shall offer the rules for comparison hi English.
(14) Of the two wars it was the first, not the second, which had produced the
greater poetry.
(iii) Certain fixed expressions: the Almighty / the accused/ the deceased/the
condemned / the departed:
(15) The accused was/were greeted by many supporters.
As we have seen, substantially used adjectives normally require the definite
article. There are, however, exceptions to this, especially in the area of fixed
expressions, e.g.: to do one's best / at last / at least / my beloved / my
intended / our poor / a bird feeding its young / a new class of rich (cf.
Schibsbye 1970: 123f). Note that in several of these a possessive pronoun
functions similarly to a definite article.
The distinction between substantival use of adjectives with specific and
with (sub)generic reference is not always clear-cut. Thus, we occasionally
come across examples with superlative forms with generic reference and,
conversely, examples of positive forms with rather specific meaning:
(16) Equally hard to handle were the aggressively overconfident. But the worst,
Maggie confided, were the conceited.
(17) Oh, come on, Leo, don't play the innocent with me!
In this last example, the innocent still denotes 'type' but clearly does not have
general plural meaning.
12.5. Adverbals
12.5.1. Preliminaries
We begin our treatment of adverbals with some remarks on adverbs as a
word class. As indicated by its name, an adverb is a word which is closely
connected with a verb. In this way adverbs differ from adjectives, which
normally attach themselves to nouns. This difference can be illustrated by an
S P A sentence like Numerous Danes drink immoderately, in which
immoderately is closely connected with drink and numerous is attached to
Danes. Now to a large extent the words that are closely connected with verbs
are the same as those that modify adjectives and different from the ones that
modify nouns (see Huddleston 1984: 330). This can be illustrated by the
following examples:
(1) She admires him excessively.
(2) The debate was excessively long.
(3) She drank an excessive amount of whisky.
The words that are closely connected with verbs and which modify
adjectives also modify words like themselves, i.e. other adverbs:
(4) She spoke excessively quickly.
The fact that words like excessively share these three properties essentially
"provides the rationale for the very broad definition of adverb given in
traditional grammar" (Huddleston 1984: 330). Some words normally
classified as adverbs, it should be added, do not have all three relational
properties (for example, NOT, ACROSS and UPSTAIRS cannot modify an
adjective or another adverb). Others modify not just the verb but the rest of
the clause:
(5) Svuprisingly/Unfortunately/Sadly, no one turned up.
As compared with the other open word classes (nouns, verbs, adjectives)
adverbs are highly heterogeneous, particularly semantically. This can be
illustrated by words like NOT, WELL, ACROSS, ELEGANTLY and CLOCK-
WISE. It is therefore not surprising that they have been defined negatively as
those words that belong to an open class other than the noun class, the verb
Adverbals 487
class or the adjective class (i.e. as the default open word class). In positive
terms, adverbs can be characterized as words whose functional domain is to
express properties in relation to situations (as expressed by a verbal or a
whole clause) or in relation to other properties (as expressed by adjectivals or
other adverbals).
Morphologically, a large number of adverbs can be identified by the suffix
-fy (for example SLOWLY, FAINTLY, REMARKABLY, BLATANTLY, CARE-
FULLY, NICELY). We return to the morphology of adverbs in section 12.5.2.
Let us now consider adverbals. As pointed out in section 3.3.1, an adverbal
is either a single adverb or a group with an adverb as head. This can be
illustrated by the following examples:
(6) Please contact us directly for brochures or subscription information.
(7) The boys are playing very nicely.
(8) She was so surprised she stopped right there.
(9) They may arrive sooner than you expect.
(10) You just didn't look carefully enough.
(11) Lack of money is wiping out Russia's armed forces more efficiently than any
nuclear bomb.
(12) She jerked her head so violently she woke the baby.
As appears, the head adverb of an adverb group accepts a pre-head
dependent (7, 8), a post-head dependent (9, 10) or a discontinuously realized
dependent which envelops the head adverb (11, 12). Occasionally an adverb
may have two dependents, as in e.g. very fortunately for me in which there is
not discontinuity but two separate dependents.
Pre-head dependents in adverbals are adverbs or adverb groups which
typically modify a gradable adverb with respect to degree. This is the case in
e.g. (7) and can be further illustrated by
(13) The rain smelled ever so slightly of home.
In both these sentences the pre-head dependent (very, ever so) intensifies the
meaning of the head adverb, and so it does in an example like (8) in spite of
the fact that THERE is not a gradable adverb, i.e. right there signals exact
location and in that sense a high degree of'thereness'. It should be noted that
a pre-head dependent may also signal a lower degree of the quality expressed
by the head adverb, cf. e.g. She behaved less politely last week. Frequent
degree adverbs found in pre-head position are the ones which also premodify
adjectives (see section 12.1.1): VERY, M U C H , QUITE, EXTREMELY,
RATHER, HIGHLY, REALLY, SO, AS, TOO, MORE, MOST, LESS, LEAST.
Pre-head adverbals may also occasionally express 'viewpoint', i.e. the respect
in which something is done, as in the (She reacted) politically correctly.
Post-head dependents are realized by than-clauses (in cases of morphological

comparison) or by the adverb ENOUGH, as in (9) and (10), respectively.
They may also be realized by preposition groups:
(14) Fortunately for me, they've already left.
Discontinuous dependents are found in cases of syntactic comparison:
(15) Pundits were hedging even more blatantly than usual.
(16) Mary danced the most elegantly of them all.
They may also be realized by constructions with as ... as ..., so ... as ..., so ...
thai-clause, less... than and too + to-infinitive clause:
(17) Mary dances as elegantly as Joan.
(18) She has never danced so elegantly as tonight.
(19) Kemp and Gore got along so famously that it looked like a fraternity picnic.
(20) Mary dances less confidently than she did last night.
(21) He is speaking too eloquently to be entirely trusted.
Adverbals express a variety of meanings (see also sections 3.2.9, 7.3.2 and
8.8 on the semantics of adverbials more generally). Some of these are
illustrated in the following examples:
(22) She was so surprised she stopped right there, (place)
(23) The economy has only now begun to emerge from four years of stagnation,
(time)
(24) You just didn't look carefully enough, (manner)
(25) I love you very much, (degree)
(26) These items are rare and therefore very expensive, (reason)
(27) Even so, what would you do? (concession)
(28) Politically, he is quite mad. (viewpoint)
As mentioned in section 9.9.1, adverbals may also express modal meanings,
cf. e.g. Evidently they have slept here (epistemic) and Hopefully this is
enough (deontic).
12.5.2. Morphology
Adverbs are morphologically simple (YET, ENOUGH, etc.), complex (BACK-
WARDS, BLATANTLY, etc.) or compound (FURTHERMORE, THERE-
ABOUTS, etc.). Complex adverbs constitute by far the largest of these
subclasses, and most of them are formed by adding ~ly to an adjective. This
adverb-forming suffix is highly productive and much more so than the
adjective-forming suffix -ty (FRIENDLY, COWARDLY, etc.). Adverbs in -ly
frequently express manner. For example, the meaning of BLATANTLY is 'in
Adverbals 489
a blatant way', and words like QUIETLY, RAPIDLY, FAINTLY, VIOLENTLY,

SILENTLY, EASILY and NICELY can also be glossed 'in an [adjective] way'.
In many other cases, however, and as demonstrated by the following
examples, the meaning of an adverb in -ly is different:
(1) He's occasionally late, (time-frequency)
(2) I haven't heard from her recently, (time-duration)
(3) There's a lot of crime locally, (place)
(4) Theoretically, this is a good solution, (viewpoint)
Note that an adverb in -ly may express manner if it functions as a pre-head
dependent in an adjective group but have another meaning if it realizes an
adverbial:
(5) The room is barely [= sparsely] furnished. S P C (manner)
(6) The room is barely [= scarcely] furnished. S P A C (degree)
(7) They can barely [= scarcely] read. S P- A -P (degree)
Complex adverbs may also be formed by adding -ly to an -ing or -ed
participle form, as in JOKINGLY and HEATEDLY, or to a noun, as in
MONTHLY and YEARLY (which are matched by identical adjectives). On the
other hand, they cannot normally be formed by adding -ly to an adjective
ending in -ly, i.e. from an adjective like FRIENDLY one cannot freely derive
*FRIENDLILY. An example like Jeremy sat in a misery of embarrassment,
sicklily smiling (quoted from Schibsbye 1970: 151) is thus exceptional.
As regards spelling, it should be mentioned that the final -y of an adjective
in more than one syllable is changed to -/- in a derived -ly adverb, as in
PRETTILY (PRETTY). Furthermore, the final -e of the adjectives DUE and
TRUE is dropped in the derived adverbs DULY, TRULY (but note that it is
retained in e.g. PALELY and SOLELY). The letter -e is also omitted in
adverbs derived from adjectives ending in -le, though here an / is dropped as
well (except in WHOLLY): SUBTLY (SUBTLE), NOBLY (NOBLE), etc.
Adverbs derived from adjectives in -ic end in -ically, as in PROBLEM-
ATICALLY (PROBLEMATIC), except for PUBLICLY and POLITICLY. Note
finally an adverb like OFFHANDEDLY, derived from the adjective
OFFHAND. In this case, it should be added, the form without a suffix may
also be an adverb, as in I can't say offhand whether it's route 66 or 69.
A large number of complex adverbs in -ly are matched by adverbs without
this suffix, for example LOUDLY by LOUD:
(8) A dog was barking loudly.
(9) Read the letter out loud.
Other examples of such adverb pairs are CLOSE/CLOSELY, DEEP/DEEPLY,

DIRECT/DIRECTLY, FLAT/FLATLY, HARD/HARDLY and HIGH/HIGHLY.
While the variants in -ly in many of these pairs are usually unmarked and the
commoner ones, the short forms are largely restricted to idiomatic usage.
Examples illustrating such usage are hold me close, deep down, due east,
take it easy, your memory is playing you false, aim high, pretty stupid, come
as quick as you can, it serves you right, live rough (i.e. without the usual
facilities), sleep tight and open your mouth wide. A detailed account of
adverbs in -ly which have matching forms without a suffix is given by van
Ek & Robat (1984: 361ff) and by Schibsbye (1970: 152ff). In order to make
sure to choose correctly between the two forms, the reader is also advised to
consult a good dictionary.
Some adverbs in -ly expressing time are matched by formally identical
adjectives, for example EARLY, HOURLY, DAILY, WEEKLY, MONTHLY
and YEARLY. While it is an adverb which occurs in a sentence like I got up
early today, the -ly form in a sentence like She's in her early thirties is an
adjective.
Complex adverbs may also consist of a noun followed by the suffix -wise.
In examples like CRABWISE and CLOCKWISE, which express manner and
direction respectively, this type of derivation is unproductive. However,
adverbs may be freely formed - particularly in ArnE - by adding -wise to a
noun for the expression of viewpoint, i.e. in the sense of'so far as [noun] is
concerned'. Examples illustrating this type of formation are FOODWISE,
DRINKWISE, MONEYWISE, BUSINESSWISE and WEATHERWISE. Some
complex adverbs end in the suffix -ward(s) (the forms without -s being
typical of AmE). As illustrated by WESTWARD(S), HOMEWARD(S), BACK-
WARD^) and FORWARD(S) such adverbs express direction. Finally, a small
group of complex adverbs begin with the prefix a-: ABROAD, APART,
ASHORE, ASIDE, ASUNDER (see Huddleston 1984: 333). This type of
adverb formation is completely unproductive.
In comparison with simple nouns, verbs and adjectives, simple adverbs
(VERY, SOON, TOO, YET, etc.) constitute a fairly small class. Some simple
adverbs, for example FAR, FAST, LITTLE, LONG, LOW, ONLY, STRAIGHT
and WELL, are matched by formally identical adjectives. This can be
illustrated by examples like the following:
(10) Let's buy low and sell high, (adverbs)
(11) Though in a low state of health, she's in high spirits, (adjectives)
As pointed out by e.g. Huddleston (1984: 333), the justification for saying
that low and high are adverbs in (10) and adjectives in (11) is that it is
usually contrasting forms we find in these two types of environment, as in
Adverbals 491
buy badly and a bad state of health. Another group of simple adverbs, for
example AFTER, BY, IN, NEAR, ON, OVER, THROUGH, UNDER and UP, are
matched by formally identical prepositions:
(12) Our relationship is over, (adverb)
(13) She put her hand over my mouth, (preposition)
As illustrated by e.g. ABOARD, BENEATH, INSIDE, OUTSIDE and
UNDERNEATH, such 'prepositional adverbs' may also be complex. Finally,
attention should be drawn to simple words like WHEN, WHERE, WHY and
HOW. Though often termed 'pro-adverbs' because they represent adverbial
preposition groups (at what time / at what place /for what reason / in what
manner), these are here classified as pronouns and have therefore already
been dealt with (in sections 11.3.2 and 11.3.3).
Compound adverbs constitute a relatively small class. A number of these
have HERE or THERE as their first element and a preposition as their last, for
example AFTER, BY, IN, OF, TO or UPON. Compounds ending in these
prepositions may also begin with WHERE, but as mentioned in section 11.3.2
WHERE is in this book analysed as a pronoun. Apart from HEREABOUTS,
THEREABOUTS and THEREFORE adverbs beginning with HERE or THERE
are formal, and so are compound adverbs like EVERMORE, FORTHWITH,
HENCEFORTH, THENCEFORTH and WELLNIGH. Other examples illus-
trating this morphological subclass of adverbs are HALFWAY, NEARBY,
OUTRIGHT, STRAIGHTFORWARD and words composed of a preposition
followed by a noun such as DOWNHILL, DOWNSTAIRS, INDOORS, OFF-
HAND and OUTDOORS. Note that apart from INDOORS and OUTDOORS
these words can only be classified as adverbs rather than adjectives by
looking at the context in which they occur (see section 3.1.4).
12.5.3. VERY vs. MUCH

Among the simple adverbs VERY and MUCH are particularly common, and
as these two words usually present difficulty to foreign learners of English,
their distribution will briefly be dealt with. MUCH is used if the adverb
functions as an adverbial, i.e. if it is closely connected with the verb. This
use of MUCH is almost always restricted to non-assertive sentences:
(1) We didn't enjoy the play much.
(2) It doesn't much matter what you say.
VERY is used before positive adjectives and adverbs:
(3) This is very useful.
(4) You'll hear from me very soon.
MUCH, however, is used in constructions with too followed by a positive

form and is sometimes preferred before positive forms with the prefix a-:
(5) I was much too fond of her.
(6) Her little brother was much afraid.
VERY is used as a pre-head dependent before an adjective or adverb in the
superlative form while MUCH is used before the combination of the definite
article and the superlative form:
(7) This is our very lowest price.
(8) Friday is the very soonest I can have it ready.
(9) This is much the noisiest place.
If the head of an adjective or adverb group is in the comparative form,
however, it is MUCH which is required:
(10) This is much better.
(11) You'll notice much sooner than you expect.
Latinate 'comparatives' like SENIOR, JUNIOR, SUPERIOR, INFERIOR, etc.
take MUCH if comparative meaning is expressed but otherwise VERY, cf:
(12) This paper is much superior to the last one you submitted.
(13) This is very inferior stuff.
Note also that in accordance with the above rules an -ed participle form
combines with MUCH if it is passive and with VERY if it is an adjectival
non-passive (see section 7.4.6):
(14) Your attitude is much disliked by your colleagues. (SP-A-P A)
(15) The case is very complicated. (S P C)
Before preposition groups we only find MUCH:
(16) He was much in love.
(17) She decided to meet with him again, much to my annoyance.
Finally it should be mentioned that MUCH is often premodified by VERY and
in some idiomatic cases obligatorily so:
(18) I like it very much.
12.5.4. The external relations of adverbals

Adverbals typically function as adverbials or as dependents. When they take
on the former function, they may realize adjuncts, disjuncts or conjuncts (cf.
section 3.2.9):
(1) She couldn't bring herself to go any closer, (adjunct)
Adverbals 493
(2) Chi's arrival went smoothly, (adjunct)

(3) Unfortunately, there has been strain over China's arms and nuclear transfers
to Iran and Pakistan, (disjunct)
(4) Quite frankly, relations are no longer entirely satisfactory, (disjunct)
(5) Your book has its flaws. Nevertheless, we're going to publish it. (conjunct)
(6) My first reaction was to say no. Since then, however, I've changed my mind,
(conjunct)
When functioning as a dependent, an adverbal most often modifies an
adjective:
(7) U.S. officials find it almost inconceivable that Beijing would jeopardize its
interests.
(8) Moscow is having unseasonably warm weather.
(9) He just isn't careful enough, (postmodification)
However, an adverbal may also modify an adverb, a nominal or a pro-
nominal:
(10) The rain smelled ever so faintly of home.
(11) Only afterward did I consider the irony of the situation.
(12) Vice President Al Gore intends to visit China early next year.
(13) He's totally dedicated to his work and will do virtually anything to kill the
rest of us.
Adverbal premodification of (pro)nominals is often performed by degree
adverbs like QUITE, RATHER, ALMOST and NEARLY, as in quite a blow,
rather a pity, almost/nearly everybody (see sections 10.3.5-6 and 11.3.4).
Adverbals may also function as dependents in preposition groups:
(14) She's just returned from abroad.
(15) Do you see anything up there?
Note finally examples like She went straight to bed and We finished well
within time where straight and well may be regarded as premodifiers in
relation to a preposition.
In compound units realizing an adverbial or a dependent, adverbals may
function as conjoints:
(16) Candido had heard over and over how they had clinics and housing and food
slips for poor Americans. (A)
(17) He'd stared so long and so hard at that strip of road that it wasn't a real place
anymore. (A)
(18) The weather is unexpectedly and unseasonably warm. (DEP)
In non-clausal utterances adverbals may directly realize different speech act
functions, such as STA (e.g. Very badly said in reply to Has she been hurt?),
QUE (e.g. Very badly? said in reaction to She's been hurt) and DIR (e.g.
Down said to a dog).
Finally it should be mentioned that adverbals may function as subjects or
objects in sentences like the following where locative or temporal conditions
are referred to:
(19) Indoors is recotnmendable at this time of day.
(20) I prefer outdoors.
(21) Later is not good enough, it has to be done now.
(22) I prefer tomorrow.
12.5.5. Comparison
Adverbs which have - or which can be interpreted as having - scalar
meaning, such as EARLY, NEAR, QUIETLY and SOON (but not e.g. YET,
NOW, CLOCKWISE), are capable of being compared. The type of compar-
ison they select is usually syntactic, but a number of monosyllabic adverbs -
all of which are matched by identical adjectives - take morphological
comparison:
(1) She couldn't bring herself to go any closer.
(2) Jane stayed the longest of them all.
(3) Speak louder but don't come any nearer.
(4) Let's see who can think of an answer quickest.
(5) You'll see me sooner than you expect.
Among the derived adverbs in -ly, EARLY is compared morphologically
while a couple of others, also matched by formally identical adjectives,
vacillate between morphological and syntactic comparison:
(6) Cole was injured earlier this season.
(7) He speaks kindlier / more kindly to his children than he used to.
(8) She is playing it poorlier / more poorly than she did yesterday.
Vacillation between morphological and syntactic comparison is also found in
the case of a simple adverb like OFTEN:
(9) I hope you'll come and see us oftener/more often next year.
For spelling and pronunciation of morphologically compared forms see
section 12.3.2.
In the large majority of cases - including virtually all the numerous
adverbs formed by adding -ly to an adjective - adverbs are compared
syntactically:
Adverbals 495
(10) That's what prominent Republicans said, on the airwaves and more
vehemently in private.
(11) Mary danced the most elegantly of them all.
The compared forms quicker/quickest and louder/loudest are often found
whether or not the positive forms of the adverbs would have lacked the -ly
suffix:
(12) You'll probably get a cab quicker by walking to Waterloo Road.
When compared forms are coordinated, morphologically compared adverbs
usually precede syntactically compared adverbs:
(13) Gore spoke longer and more eloquently than Kemp.
Comparison to a lower and to the same rank on a property scale can be
illustrated by examples such as:
(14) I come here less often than I used to.
(15) Let's hear if you can play it as convincingly as you claim.
A small group of adverbs have irregular comparison (see section 12.3.3 on
irregularly compared adjectives):
badly worse worst
far farther farthest
further furthest
little less least
much more most
well better best
This manner of comparison can be illustrated by the following examples:
(16) All this happened when we least expected it.
(17) I know their secrets better than anyone else.
For the use of compared forms see section 12.3.5.
References
Aarts, Bas - Charles F. Meyer (eds.)

1995 The verb in contemporary English Cambridge: Cambridge University
Press.
Bache, Carl
1978 The order of premodifying adjectives in present-day English. Odense:
Odense University Press.
Bache, Carl
1985a Verbal aspect: a general theory and its application to present-day English.
Odense: Odense University Press.
Bache, Carl
1985b "The semantics of grammatical categories: a dialectical approach", Journal
of Linguistics 21: 51-77.
Bache, Carl
1986 'Tense and aspect in fiction", Journal of Literary Semantics 15: 66-70.
Bache, Carl
1994 "Verbal categories, form-meaning relationships and the perfect", in: Carl
Bache - Hans Basbell - Carl-Erik Lindberg (eds.), 43-60.
Bache, Carl
1995 The study of aspect, tense and action. Frankfurt am Main: Peter Lang.
Bache, Carl - Hans Basb011 - Carl-Erik Lindberg (eds.)
1994 Tense, aspect and action: empirical and theoretical contributions to lan-
guage typology. Berlin and New York: Mouton de Gruyter.
Bache, Carl - Mike Davenport - John Dienhart - Fritz Larsen
1993 An introduction to English sentence analysis. Copenhagen: Munksgaard.
Bache, Carl - Leif Kvistgaard Jakobsen
1980 "On the distinction between restrictive and nonrestrictive relative clauses in
modem English", Lingua 52: 243-267.
Bolinger, Dwight
1952 "Linear modification", Publications of the Modern Language Association
of 'America 67: 1117-1141.
Bolinger, Dwight
1967 "Adjectives in English: attribution and predication", Lingua 18: 1-34.
Bolinger, Dwight
1975 Aspects of Language (2nd edition). New York: Harcourt Brace Jovanovich,
Inc.
Chomsky, Noam
1957 Syntactic structures. The Hague: Mouton & Co.
498 References
Collins Cobuild English grammar:

see John Sinclair 1990.
Davidsen-Nielsen, Niels
1985 "Has English a future?", Acta Linguistica Hafniensia 21,1: 5-20.
Davidsen-Nielsen, Niels
1990 Tense and mood in English: a comparison with Danish. Berlin and New
York: Mouton de Gruyter.
Davidsen-Nielsen, Niels (ed.)
1991 LSP: nine studies on language for special purposes. Copenhagen: Handels-
hojskolens Forlag.
Dienhart, John
1992 "Adverbials, direct objects and the style of Carson McCullers", The Twain
Shall Meet, POET 18, Department of English, University of Copenhagen:
121-134.
Dik, Simon
1989 The theory of functional grammar. Dordrecht: Foris Publications.
Dirven, Reno (ed.)
1989 A user's grammar of English: word, sentence, text, interaction. Frankfurt
am Main: Peter Lang.
Downing, Angela - Philip Locke
1992 A university course in English grammar. New York and London: Prentice
Hall.
Elsness, Johan
1997 The perfect and the preterite in contemporary and earlier English. Berlin
and New York: Mouton de Gruyter.
Ferris, Connor
1993 The meaning of syntax: a study in the adjectives of English. London: Long-
man.
Fodor, Jerry A. - Jerrold J. Katz (eds.)
1964 The structure of language. Englewood Cliffs, N. J.: Prentice-Hall.
Granger, Sylviane
1983 The be + past participle construction in spoken English with special em-
phasis on the passive. Amsterdam: North Holland.
Greenbaum, Sidney - Randolph Quirk
1990 A student's grammar of the English language. London: Longman.
Halliday, Michael A. K.
1985 Spoken and written language. Victoria: Deakin University Press.
Halliday, Michael A. K.
1994 An introduction to functional grammar. London, New York, Sydney and
Auckland: Edward Arnold.
References 499
Hansen, Erik
ms. "£>ef-kl0vning".
Harder, Peter
1989 "The instructional semantics of conditionals", Working Papers in Func-
tional Grammar 30: 1-42.
Harder, Peter
1996 Functional semantics: a theory of meaning, structure and tense in English.
Berlin and New York: Mouton de Gruyter.
Hartvigson, Hans - Leif Kvistgaard Jakobsen
1974 Inversion in present-day English. Odense: Odense University Press.
Hopper, Paul J. - Elizabeth C. Traugott
1993 Grammaticalization. Cambridge: Cambridge University Press.
Hoye, Leo
1997 Adverbs and modality in English. London: Longman.
Huddleston, Rodney
1976 "Some theoretical issues in the description of the English verb", Lingua 40:
331-383.
Huddleston, Rodney
1984 Introduction to the grammar of English. Cambridge: Cambridge University
Press.
Huddleston, Rodney
1988 English grammar: an outline. Cambridge: Cambridge University Press.
Huddleston, Rodney
1995 "The English perfect as a secondary past time", in Bas Aarts - Charles F.
Meyer (eds.), 102-122.
Jakobsen, Leif Kvistgaard
1994 "Variation in the use of the definite article and the demonstratives as
cohesive devices", RASK 1 (Odense University Press): 63-82.
Jespersen, Otto
1909-49 A modern English grammar on historical principles. Copenhagen: Munks-
gaard, and London: Allen & Unwin.
Jespersen, Otto
1929 The philosophy of grammar. London: Allen & Unwin.
Jespersen, Otto
1933 Essentials of English grammar. London, Boston and Sydney: George Allen
& Unwin.
Johansson, Stig - Per Lysväg
1986 Understanding English Grammar 1. Oslo: Universitetsforlaget.
Johansson, Stig - Per Lysväg
1987 Understanding English Grammar 2. Oslo: Universitetsforlaget.
500 References
Juul, Arne
1975 On concord of number in modern English. Copenhagen: Nova.
Juul, Ame - Knud Serensen
1978 Numerus i moderne engelsk. Copenhagen: Sch0nberg.
Klima, Edward
1964 "Negation in English", in: Jerry A. Fodor - Jerrold J. Katz (eds.), 246-323.
Klinge, Alex
1993 "The English modal auxiliaries: from lexical semantics to utterance inter-
pretation", Journal of Linguistics 29: 315-357.
Kragh, Bodil
1991 "LSP, science and technology: a sociological approach", in: Niels David-
sen-Nielsen (ed.), 39-52.
Lakoff, George - Mark Johnson
1980 Metaphors we live by. Chicago and London: University of Chicago Press.
Lauridsen, Karen
1986 Ovelser til engelsk grammatik. Copenhagen: Sch0nberg.
Leech, Geoffrey
1981 Semantics (2nd edition). Harmondsworth: Penguin.
Lyons, John
1968 Introduction to theoretical linguistics. Cambridge: Cambridge University
Press.
Lyons, John
1981 Language and linguistics. Cambridge: Cambridge University Press.
Lyons, John
1995 Linguistic semantics: an introduction. Cambridge: Cambridge University
Press.
Martin, James Edward
1968 A study of the determinants of preferred adjective order in English. Un-
published doctoral dissertation, Urbana, Illinois: University of Illinois.
Matthews, Peter
1981 Syntax. Cambridge: Cambridge University Press.
McArthur, Tom (ed.)
1992 The Oxford companion to the English language. Oxford: Oxford University
Press.
Munck, Lena
1991 "Les textes techniques: leurs elements constitutifs et leur realisation en da-
nois", in: Niels Davidsen-Nielsen (ed.), 53-66.
Nässlin, Siv
1984 The English tag question. Stockholm: Almquist & Wiksell.
References 501
Nelke, Henning
1984 "Clefting in Danish", Nydanske Studier & Almen Kommunikationsteori 14
(Copenhagen: Akademisk Forlag): 72-111.
Palmer, Frank
1986 Mood and modality. Cambridge: Cambridge University Press.
Palmer, Frank
1987 The English verb (2nd edition). London: Longman.
Palmer, Frank
1990 Modality and the English modals (2nd edition). London: Longman.
Payne, John
1985 "Negation", in Timothy Shopen (ed.), 197-242.
Preisler, Bent
1992 A handbook of English grammar. Aarhus: Aarhus University Press.
Quirk, Randolph - Sidney Greenbaum - Geoffrey Leech - Jan Svartvik
1985 A comprehensive grammar of the English language. London: Longman.
Radden, Günter
1989 "Semantic roles", in: Rene Dirven (ed.), 421 -472.
Schibsbye, Knud
1970 A modern English grammar (2nd edition). Oxford: Oxford University
Press.
Shopen, Timothy (ed.)
1985 Language typology and syntactic description I: clause structure. Cam-
bridge: Cambridge University Press.
Sinclair, John (ed.)
1990 Collins Cobuild English grammar. London and Glasgow: Collins.
Steller, Poul - Knud S0rensen
1974 Engelsk grammatik (2nd edition). Copenhagen: Munksgaard.
Swan, Michael
1995 Practical English usage. (2nd edition). Oxford: Oxford University Press.
Twaddell, W. Freeman
1965 The English verb auxiliaries. Providence: Brown University Press,
van Ek, Jan A. - Nico J. Robat
1984 The student 's grammar of English. Oxford: Basil Blackwell.
Vestergaard, Torben
1985 Engelsk grammatik. Copenhagen. Schanberg.
Vikner, Carl
1973 "Quelques reflexions sur les phrases clivees en frangais moderne", Actes du
Seme Congres des romanistes scandinaves, Turku, 221-235.
Wells, Rulon S.
1947 "Immediate constituents", Language 23:81-117.
Subject index
Ability 9.9.1 - Distributive sequence 12.1.5

Absolute clause 8.8 - Gradable 12.3.5
Absolute sense - Irregular comparison 12.3.3
- of adjectives 12.3.5 - Peripheral 12.1.6
Accusative case 11.2.1 [D] - Positional restrictions 12.1.7
Accusative with infinitive 5.6.1 - Predicative 12.1.4
Acrolect 1.6 - Semantics 12.1.2
Action category 7.2.1 - Specifying 12.1.6
Actionality 7.2.1 - Substantival use 12.4
- and progressive forms 9.6.6 - Superlative 11.3.4[E]
Active voice 3.2.5, 7.4.1 - and attributive position 12.1.7
Active without passive 7.4.4 [B] - and end-focus 12.1.7
Address 4.5 - and morphology 12.1.3
Adjectival 3.3.1, 12 - and predicative position 12.1.7
- Absolute sense 12.3.5 - and ranking 12.3.5
- Appositional 12.1.4 - in indefinite pronoun group 12.1.7
- External relations 12.1.4 - vs. adverb 12.5.2
- Functional domain 12.2.1 Adjective group 3.3.1; 3.2.7-8,12.1.1
- Non-restrictive 12.1.4 - Discontinuous 5.6.5
- Restrictive 12.1.4 - and post-head dependent 12.1.1
Adjectival comparison 12.3 - and pre-head dependent 12.1.1
- Morphological 12.3.1 Adjunct 3.2.9,4.1.2, 5.5.10-13, 8.1,
- Syntactic 12.3.1 12.5.4
Adjectival Mod. zone 12.2 - Relative 11.3.3 [B.c]
- Structure 12.2.4 Adverb 3.1.4, 3.2.9,4.2.3,4.3.1-4,
- Internal order 12.2.5 [A]-[C] 11.1.2,12.5.1-5
Adjectival modification 12.2 - Complex 12.5.2
Adjectival non-passives 7.4.6 - Compound 12.5.2
Adjectival past participle 12.1.3 - Irregular comparison 12.5.5
Adjectival subfunctions - -ly adverb 12.5.2
- Classification 12.2.2 - Morphology 12.5.2
- Description 12.2.2 - Simple 12.5.2
- Specification 12.2.2 - as post-head dependent 12.5.1
Adjective 3.1.4, 3.2.7-8, 7.4.6, 12.1-4 - as pre-head dependent 12.1.1, 12.5.1
- Attributive 12.1.4 - of degree 10.3.5
- Central 12.1.6 - particles in phrasal verbs 4.3.1
- Classifying 12.1.6 - vs. adjective 12.5.2
- Comparative 11.3.4 [E] - vs. preposition 12.5.2
- Comparison 12.3 Adverb group 3.3.1
- Compound 12.1.3 - Discontinuous 5.6.5
- Denominal 12.1.3, 12.2.5 [B] - as pre-head dependent 12.1.1, 12.5.1
- Descriptive 12.1.6 Adverbal 12.5
- Deverbal 12.1.3, 12.1.7, 12.2.5 [B] - External relations 12.5.4
- Dislocated 12.1.4 - Modification 12.5.3
504 Subject index
- as pre-head dependent 12.1.1 - Direct 10.3.4

- as post-head dependent 12.1.1 - Extrasentential 10.3.4
Adverbal comparison 12.5.5 - Indirect 10.3.4
- Irregular 12.5.5 - Intrasentential 10.3.4
- Morphological 12.5.5 - Non-repetitive 10.3.4
- Syntactic 12.5.5 - Non-textual 10.3.4
Adverbial 3.2.9; 3.2.3,4.3.2,4.5, 7.4.5, - Repetitive 10.3.4
7.5.8, 8.8,9.6.4,12.5.4 - Textual 10.3.4
- Concessive 8.9 - and demonstrative pronoun 11.3.1 [D]
- Fronting 7.5.6,1.7 - and relative pronoun 11.3.3 [B.c]
- Initial position 5.5.2 - and sentential relative clause 8.11
- Medial position 5.5.3 Antecedent 10.3.4
- Optional 7.3.4, 8.1 Apodosis 8.9
- Position and order 5.5. Appended coordination 6.2.3
- Position and scope 5.5.10, 5.5.13 Apposition 5.6.3,10.1.3, 12.1.4
- Position and semantic clarity 5.5.11 Arabic 1.7, 5.1
- Position and style 5.5.12 Article 3.1.4
- Position of infiniteclauses 5.5.5 - Definite 10.3.1,10.3.5
- Position of in nonfmite clauses 5.5.6 - Indefinite 10.3.1,10.3.6
- Semantic function 8.8 - and possessive pronoun 11.2.2 [C]
- Terminal position 5.5.4 Aspect 9.6,9.6.6
- Unattached 8.8 - and progressive form 9.6.6
- and choice of relative pronoun Assertive forms 7.5.7
11.3.3[C.e] - and indefinite pronouns 11.3.4 [A.fJ
Adverbial anchor 9.6.2 Attitudes (stative situation) 7.2.3, 7.4.3
Adverbial clauses 6.4.1, 8.8 [A]
- Semantic functions 8.8 Attraction 7.6.2, 7.6.3 [B], 7.6.5
Adverbial demonstratives 11.3.1 Attraction inversion 5.3.4
Adverbial subfunctions ATTRIBUTE (Specific participant
- Adjunct 3.2.9,4.1.2,5.5.10-13, 8.1, role) 7.3.2^
12.5.4,12.5.4 Attributive adjective 12.1.4
- Conjunct 3.2.9,4.1.2, 5.5.9-13, 12.5.4 Attributive positions
- Disjunct 3.2.9,4.1.2, 5.5.10-13, 8.1, - of adjectives 12.1.7
8.8, 12.5.4 Attributive use
- Optional 7.3.4, 8.1 - of adjective 12.1.4
AFFECTED (Specific participant role) Australian English 1.4, 1.6
3.2.5, 7.3.2-4 Autonomous genitive 10.3.8 [A],
AGENT (Specific participant role)
3.2.5, 7.3.2-4, 8.3 Autonomous use
Agent öy-group 3.2.5 - of demonstrative pronoun 11.3.1
Agglomerese 1.8 - of dual pronoun 11.3.4 [D]
Aggregate noun 10.4.4 - of emphatic indefinite article 11.3.4
Ambiguity 2.6 [E]
American English 1.5-6, 6.4.1, 7.4.5, - of indefinite pronoun 11.3.4 [A.a]
7.5.2,7.6.4,9.2,9.8.3,12.5.2 - of interrogative pronoun 11.3.2 [A]
Anaphoric ellipsis 4.2.2 - of possessive pronoun 11.2.2
Anaphoric reference 8.10 [C], 10.3.4, - of pronoun 11.1.3
- of relative pronoun 11.3.3 [A]
Subject index 505
- of universal pronoun 11.3.4 [B-C] - Intrasentential 10.3.4

Auxiliary verb 3.2.1,4.3.1 9.1 - Non-repetitive 10.3.4
- Central 9.5 - Non-textual 10.3.4
- Delimitation 9.5 - Textual 10.3.4
- Modal verb 9.5 - and demonstrative pronoun 11.3.1
- Order 9.4 [D]
- Primary verb 9.5 - and relative pronoun 11.3.3 [B.c]
Auxiliary-lexical scale 9.5 Categorical utterance 9.9.1
Categorization 10.2; 10.1.4,11.1.4
Backshifting - Class-member 10.2.4
- and tense-aspect forms 9.6.13 - Generic 10.2.4
- and past perfect 9.6.4 - Unique 10.2.4
- and present future progressive 9.6.8 Catenative verbs 7.4.5,9.5
Bare infinitive clause 8.2, 8.6 Causal linking 8.9
Base form 9.2 CAUSE (Specific participant role)
Basic sentence patterns 3.2.3, 7.1, 7.3.2^
7.3.4,8.1 Central adjective 12.1.6
Basic sentence structures 3.2.3 Central determination 10.3.2
Basilect 1.6 Central pronoun 11.2; 11.1.2
BENEFICIARY (Specific participant Channels] of communication 1.7
role) 3.2.6,7.3.2-4 Chicano English 1.6
Binary coordination 6.1 Choice of relative pronoun 11.3.3
Binary group 3.3.2 [C]
Binary noun 10.4.3 - after a preposition 11.3.3 [C.a]
Binary subordination 6.1 - as adverbial 11.3.3[C.e]
Biographical narrative - as complement 11.3.3 [C.b]
- and tense-aspect forms 9.6.14 - as object 11.3.3 [C.d]
Black English 1.4, 1.6 - as subject 11.3.3 [C.c]
Block language 4.4.5 - in discontinuous preposition group
Bound morphemes 2.1 11.3.3 [C.d]
British English 1.4-7, 5.3.2, 5.3.10, Choice relation 2.8
6.4.1,7.4.5,7.5.2, 7.6.3-4,9.2, 9.8.3, Chomskyan Generative Grammar 1.1,
11.3.3 [B.a], 12.3.1 2.6
Äy-group 7.4.1, 7.4.4 [A], 7.4.6, 8.3 CIRCUMSTANCE (Specific partici-
pant role) 7.3.2-4
Canadian English 1.6 Class-member categorization 10.2.4
Cardinal numbers 10.3.3 Class-member description 10.3.6 [B]
Caribbean English 1.6 Class-member reference 10.3.6 [A]
Case - Indefinite 10.3.7 [A]
- and indefinite pronouns 11.3.4 [A.c] Class-member referent 10.2.4
- and interrogative pronouns 11.3.2 Classification
[B.a] - Adjectival subrunction 12.2.2
- and pronouns 11.2.1 [D] Classification of pronoun 11.1.2
- and relative pronouns 11.3.3 [B.b] Classifying adjective 12.1.6
Cataphoric ellipsis 4.2.2 Classifying genitive 10.3.8 [A], 12.2.5
Cataphoric reference 8.10 [C], 10.3.4, [C]
11.2.1[F] Clausal complementation 8.5
- Extrasentential 10.3.4 Clausal subordination 6.4
506 Subject index
Clause 2.2, 3.1.3, 3.3.4,4.1.1,4.4.3 Closed condition 8.9

- Absolute 8.8 Closed word class 3.1.4
- Adverbial 6.4.1 Co-occurring determiners 10.3.2
- Comment 8.10 [A] Cockney 1.4,1.6
- Comparative 9.8.3 Code-construction 9.5
- Concessive 5.4.4,9.8.3 Cognate object 7.4.4 [B]
- Conditional 8.9,9.8.3 Cohesion
- Conditional-concessive 6.4.4 - Contextual 5.5.9
- Correlative that- 5.3.4 - Textual 5.5.9
- Counterfactual 8.9 Collective noun 7.6.4 [A], 10.4.4
- Declarative 4.2.2 Collectivizing
- Discontinuous 5.6.1 - of noun 10.4.6
- Discontinuous subject 11.2.1 [D] Collocation of adverb and modal verb
- Exclamatory 4.4.4 9.9.1
- Finite 5.4.4 Colon convention 3.1.3
- J/-8.9 Comment & topic 3.2.2,4.1.2
- Imperative 4.4.4 Comment (rheme) 5.2.3
- Indirect object 8.7 Comment clause 1.7, 8.10 [A]
- Interrogative 4.4.4 Commitment 9.9.5
- Interrogative subclause 11.3.2 [A] Common noun 10.2.4
- Matrix 6.4.1,9.6.3,9.6.7 Communicative competence 2.7
- Nominal 8.2 Communicative functions 2.5,4.4.2-3
- Non-declarative 4.4.4 Communicative grammar 1.2
- Non-restrictive relative 8.4,11.3.3 [A] Comparative
- Nonfinite 5.4.4 - Meaning 12.3.1
- Nonfinite subclause 11.2.3 [A] - and definite article 12.1.1
- Object complement 8.7 Comparative adjective 11.3.4 [E]
- Reporting 5.3.9 Comparative basis 5.6.5,12.3.1
- Restrictive relative 8.4,10.3.5 [A], Comparative clause 9.8.3
11.3.3 [A] Comparative constructions 5.6.5
- Sentential relative 8.10 [C], 11.3.3 Comparative rank 5.6.5,12.3.1
[B.c] - Positive 12.3.1
- Subject complement 8.6 - Comparative 12.3.1
- Subject 7.6.3 [C], 8.3 - Superlative 12.3.1
- Subjectless 8.2 Comparative element 5.6.5,12.3.1
- Supplement!ve 8.8 Comparative form 12.3.5
- Tag8.10[B] Compared forms
- that- 8.2-3, 8.6 - Use 12.3.5
-to-infinitive5.6.1 Comparison 5.6.5
- Verbless 5.4.4 - Analytic 5.6.5
- wh-interrogative 4.4.4, 8.2-3 - Inflectional 5.6.5
- yes-no interrogative 4.4.4, 8.2-3 - Irregular 12.3.3
Clause-initial position 11.3.3 [A] - Morphological 5.6.5,12.3.1
Cleft sentence 7.5.1,8.4 - Syntactic 5.6.5, 12.3.1
- Pseudo8.4 - Synthetic 5.6.5
- Reduced 8.4 - of adjective 12.3.1
— Restrictions 8.4 - of adjective, spelling and pronuncia-
— Syntactic property 8.4 tion 12.3.2
Subject index 507
- of adverb 12.5.5 - Inferential 8.9

Competence 2.7 - Interrogative 8.9
Complement 3.1.2 Conditional Utterance 1.7
- and choice of determiner 10.3.6 [B], Conditional-concessive clause 6.4.4
10.3.7 [E] Conjoints 3.3.3,6.1
- and choice of relative pronoun Conjunct
11.3.3[C.b] - 3.2.9,4.1.2,5.5.9-13,12.5.4
- to the object 3.2.4, 3.2.8 Conjunctions 3.1.4,3.3.2-3,6.1
- to the subject 3.2.3-4,3.2.7,11.2.1 - Complex subordinating 6.4.3
[D] - Coordinating 6.2.4
Complementation 6.3.3 [B], 10.1.4, - Correlative 6.2.4
12.1.1 - Subordinating 6.4.1
Complex adverb 12.5.2 Consecutio temporum principle
Complex coordination 4.1.2, 6.2.3 9.6.13
Complex lexicalization 6.3.3 [D] Consist-of relationship 3.1.2
Complex predicators 4.3.1-4,6.3.3 [D] Consonant doubling 9.2
Complex sentence 7.1, 8.1-12 Constituency 2.5
- Polarity 8.11 Constituent 2.5
Complex subordinating conjunctions - Initial and partial inversion 5.3.4
6.4.3 - and inversion 5.3.10
Complex-transitivity 3.2.4, 3.2.6 Constituent order 5.1
Compound 10.2.1 - in subordinate clauses 5.4
Compound adjective 12.1.3 Constraints on the passive 7.4.4 [B]
Compound adverb 12.5.2 Context
Compound unit 3.1.3,3.3.3,4.4.3, — Extralinguistic 9.1
7.6.3 [B], 7.6.4 [E], 12.1.5 - Linguistic 7.4.3
Compound word 2.1 - Situational 7.4.3
Compulsion Contraction 7.5.2
- and deontic subjunctive 9.8.3 - NOT-7.5.2
- and modality 9.9.3-5 - Verb 7.5.2
CON (Condition) 4.4.2 Contrast-formation 10.1.4
Concessive (deontic) subjunctive 9.8.3 - Non-restrictive 10.1.4
Concessive clause 5.4.4, 9.8.3 - Restrictive 10.1.4
Concord 3.2.2, 7.6.1-2 Contrastive focus 7.5.10
- External 7.6-7 CONTROLLER (Specific participant
- Internal 7.6 role) 7.3.2^
- Negative 7.5.5 Coordinated predicate 4.1.2
- Object and object complement 7.7 Coordinating conjunctions 6.2.4
- Subject and complement 7.7 Coordination 3.3.3, 6.2, 6.1
- Subject-predicator 7.6 - Appended 6.2.3
Condition 4.4.2 - Binary 6.1
- Closed 8.9 - Complex 6.2.3
- Open 8.9 - Irreversible 6.2.3
Conditional clause 8.9, 9.8.3 - Linked 3.3.3, 6.2.3, 12.1.5
Conditional commitment 5.3.3 - Mixed 6.2.2
Conditional sentence 9.6.3 - Multiple 6.1
- Exclamatory 8.9 - Non-recursive 6.2.3
- Imperative 8.9 - Pseudo-6.2.3
508 Subject index
- Recursive 6.1,6.2.3 - Simple past 9.6.2

- Reversible 6.2.3 Deictic nouns 10.3.5 [D]
- Simple 6.2.3 Deixis 9.6.1
- Types 6.2.3 - Loss of temporal... 9.6.14
- Unlinked 3.3.3, 6.2.3 - and demonstrative pronoun 11.3.1
- of predicates 4.1.2 [C], [E]
Coordinators 3.3.3 Delimitation of auxiliaries 9.5
Copula verbs 3.2.7; 3.2.3 Demonstrative pronoun 3.1.4,11.3.1;
Coreference 7.4.4 [B], 8.10 [B], 11.2.3 11.1.2
[A] - Autonomous use 11.3.1
Corpora 1.3 - Deixis and... 11.3.1 [C], [E]
Corrective focus 7.5.10 - Determinative use 11.3.1
Correlative conjunction 6.2.4, 7.6.3 [B] - Determiner function 10.3.1
Correlative subordinators 6.4.4 - Distant pronoun 11.3.1 [D]
Correlative thai-clause 5.3.4 - Near pronoun 11.3.1 [D]
Count noun 10.2.5,10.3.5 [A], 10.3.6 - Non-deictic use 11.3.1 [E]
[E], 10.3.7 [A], 10.3.7 [E] - and anaphoric reference 11.3.1
Countability 10.2.5 [D]
- and names 10.2.5 - and cataphoric reference 11.3.1 [D]
Countable quantification 10.1.4 - and number 11.3.1 [A]
Counterfactual clause 8.9 - and sentential relative clause 8.10
- reference to persons 11.3.1 [B]
Dangling participles 8.8 Denial 7.5.1,7.5.5
Danish 2.1, 7.5.5,12.1.7, 12.3.5 Denominal adjective 12.1.3,12.2.5 [B]
Data 1.3 Deontic modality 9.7, 9.9
Declarative sentence 4.4.2-3, 5.1, 5.2.2, Deontic subjunctive
5.6.4 - and compulsion 9.8.3
Default role 7.3.2 - and wishes 9.8.3
Default sentence function 3.2.9 Deontic use
Definite article 10.3.5; - of future forms 9.7
- Determiner function 10.3.1 - of imperative mood 9.8.1
- Emphatic use 10.3.5 [E] - of subjunctive mood 9.8.1,9.8.3
- and comparative 12.1.1 Dependency grammar 6.3.2
- and possessive pronoun 11.2.2 [C] Dependent 3.3.1
Definite determination 10.1.4, 10.3.1, - Post-head 10.1.3
10.3.8 [A] - Pre-head 10.1.3
Definite non-specific reference 10.3.5 Description (Adjectival subfunction)
[B] 12.2.2
Definite specific reference 10.3.5 [A] Descriptive adjective 12.1.6
Degree adverb 10.3.5 Descriptive modification 10.1.4
Deictic cline 9.6.12 Desirability
Deictic [verb] forms 9.6.1,9.6.2 - and modality 9.7
- Dramatic present 9.6.2 Determination 10.3; 6.3.3 [A], 10.1.4,
- Historic present 9.6.2 11.1.4
- Inclusive present 9.6.2 - Central 10.3.2
- Instantaneous present 9.6.2 - Definite 10.1.4,10.3.1,10.3.8 [A]
- Past 9.6.2 - Indefinite 10.1.4, 10.3.1
- Present 9.6.2 - Interrogative 10.3.1
Subject index 509
- Post-10.3.2 Discontinuous preposition groups 5.6.4

- Pre-10.3.2 - and choice of relative pronoun
- Relative 10.3.1 11.3.3[C.d]
- Unmarked 10.3.6 [A] Discontinuous pronoun group 5.6.3
- Zero 10.3.7 Discontinuous subject clauses 5.6.1,
- and indefinite pronouns 11.3.4 [A.a] 7.4.4 [A], 11.2.1 [D]
- and quantification 10.3.3 Discontinuous verb group 5.6.2
Determinative use Disjunct 3.2.9,4.1.2, 5.5.10-13, 8.1,
- of demonstrative pronoun 11.3.1 8.8, 12.5.4
- of dual pronoun 11.3.4 [D] - Clausally realized 8.10
- of emphatic indefinite article 11.3.4 Dislocation 1.7,4.5
[E] - Left 4.5
- of indefinite pronoun 11.3.4 [A.a] - Right 4.5
- of interrogative pronoun 11.3.2 [A] - of adjective 12.1.4
— of possessive pronoun 11.2.2 - of noun groups 4.5
- of pronoun 11.1.3 Distant pronoun 11.3.1 [D]
- of relative pronoun 11.3.3 [A] Distributive plural 7.7,10.4.6
- of universal pronoun 11.3.4 [B-C] Ditransitive prepositional predicator
Determiner function 4.3.2
- of definite article 10.3.1 Ditransitivity 3.2.4, 3.2.6,4.3.2-3
— of demonstrative pronoun 10.3.1 DO-support 3.2.1, 5.3.2,7.5.2, 9.5
- of indefinite article 10.3.1 - for emphasis 7.5.10
- of independent relative pronoun DOER (General participant role) 7.3.1-4
10.3.1 DOER 10.3.8 [B]
- of interrogative pronoun 10.3.1 - and imperative 9.8.2
- of possessive pronoun 10.3.1 DOER avoidance 7.4.2
- of relative pronoun 10.3.1 DOER Ay-group 7.4.1
Determiners 6.3.3 [A] Domain of negation 7.5.3
- Possible collocations 10.3.2 DONE-TO (General participant role)
Deverbal adjective 12.1.3,12.1.7, 7.3.1-4
12.2.5 [B] DONE-TO 10.3.8 [B]
Diglossia 1.7 - and imperative 9.8.2
Dimensions of language 2.8 Double negation 7.5.5
DIR [directive] 4.4.2 Doubling of consonants 9.2
Direct object 3.2.4-5 Dramatic present 9.6.2
Directed [dynamic] situation 7.2.2 Dual pronoun
Directed situations - Autonomous use 11.3.4 [D]
- and progressive form 9.6.6 - Determinative use 11.3.4 [D]
Directive 9.8.2 Dutch 5.3.5
- Indirect 8.1 Dynamic 3.2.8
Discontinuity 2.5, 3.1.6, 5.6 Dynamic adjective 12.1.2
Discontinuity [Internal...] 5.6.3, 5.6.5 Dynamic situations 7.2.1
Discontinuous adjective group 5.6.5 - Directed 7.2.2
Discontinuous adverb group 5.6.5 - Iterative 7.2.2
Discontinuous clauses 5.6.1 - Punctual 7.2.2
Discontinuous group 3.3.2 - Self-contained 7.2.2
Discontinuous noun group 5.6.3 - Telic 7.2.2
Discontinuous object clause 5.6.1 Dynamic verbs 3.2.7-8
510 Subject index
Dynamic/Stative distinction 3.2.7-8, Exclamatory conditional sentence 8.9

7.3.2 Exclamatory sentence 4.4.2-4, 5.2.2,
7.5.9
Ease of articulation 9.2 Existential sentence 3.2.2
Echo-question 11.3.2 Existential there 3.2.2, 5.3.7
-ed form 9.2 Extensive relations (stative situation)
-ed participle clause 8.2, 8.8 7.2.3
Edited speech 1.7 External concord 7.6-7
EGP (English for General Purposes) External focus
1.8 - and progressive form 9.6.6
Elaboratives 10.1.3 External relations
Elicitation 1.3 - of adjectivals 12.1.4
Ellipsis 4.2.1-2, 6.2.1,6.2.4,12.4.1 - of adverbals 12.5.4
- Anaphoric 4.2.2 - of pronominals 11.1.3
- Cataphoric 4.2.2 - of verbals 9.3
- Extrasentential 4.2.2 EXTRA (General participant role)
- Initial 4.2.2 7.3.1-4
- Intrasentential 4.2.2 Extralinguistic context 9.1
- Medial 4.2.2 Extraposed real object 3.2.5
- Nontextual 4.2.2 Extraposition 3.2.2, 5.2.3,5.6.1,8.3
- Terminal 4.2.2 Extrasentential ellipsis 4.2.2
Elliptical clauses 5.5.3
Elliptical constructions 1.7 Fact-mood 9.8.1
Embedding 3.3.5, 5.6.1 Finite clause 3.2.1, 5.4.4, 8.2-3, 8.5-6,
Emphasis 7.5.10 8.8
Emphatic dependent use Finite predicator 11.2.1 [D]
— of reflexive pronoun 11.2.3 [B] Finiteness 3.2.1
Emphatic autonomous use Penalization 5.2.3,5.3.4, 8.4
- of reflexive pronoun 11.2.3 [C] Focus 5.2.3, 7.5.10
Emphatic definite article 10.3.5 [D] - Contrastive7.5.10
Emphatic imperative 9.8.2 - Corrective 7.5.10
Emphatic use - Double 5.3.4
- of definite article 10.3.5 [E] Focus of negation 7.5.1
End-focus 7.4.2, 8.4 ybr-group 8.3, 8.6
- and adjectives 12.1.7 Form & function 3.1.1
End-weight 5.2.3, 5.3.9, 5.3.10, 5.5.8, Form stack 4.1.1-2
5.6.1,5.3.9-10,6.2.3,7.4.2 Form types 3.1.3
Entailment 7.5.8 Formal vs. informal 1.7
Epic preterite 9.6.14 Fractions 10.3.3
Epistemic modality 9.7, 9.9 Free indirect speech
Epistemic use - and tense-aspect forms 9.6.13
- of future forms 9.7 Free morphemes 2.1,2.2
- of subjunctive mood 9.8.1, 9.8.3 French 5.3.3, 7.6.1, 8.10 [B], 9.6.3
ESP (English for Special Purposes) 1.8 Fronted adverbial and full inversion
Everyday narrative 5.3.6
- and tense-aspect forms 9.6.14 Fronted object and full inversion 5.3.9
EXC (exclamation) 4.4.2 Fronted subject complement and full
Exclamatory clauses 4.4.4, 5.4.3 inversion 5.3.8
Subject index 511
Fronting 5.1,5.2.3 - and interrogative pronouns 11.3.2

Fronting of adverbial 7.5.6 [B.b]
Full inversion 5.3.2,5.3.5,11.2.1 [E] - and personal prounouns 11.2.1 [C]
- after fronted adverbial 5.3.6 - and relative pronouns 11.3.3 [B.a]
- andfrontedobject 5.3.9 - of noun 10.2.3
- andfrontedsubject complement General English 1.5
5.3.8 General participant roles 7.3.1
- and information structure 5.3.5 General predictability 9.7
- and predication stacks 5.3.7, 5.6.2 Generative Grammar 1.1
Full verbs 3.2.1, Generic categorization 10.2.4
Function Generic reference 10.3.5 [C], 10.3.7
- Syntactic... of genitive 10.3.8 [A] [C], 12.4.2
Function stack 4.1.2-3 - and definite article 10.3.5 [C]
Function words 3.1.4 - and plural 10.3.5 [C]
Function-over-form convention 3.1.3, Generic referent 10.2.4
3.3.1 Generic-like expressions 10.3.6 [D],
Functional domain 10.3.7 [D]
- of adjectivals 12.2.1 Generic-like use
- of nominate 10.1.4 - of personal pronoun 11.2.1 [F]
- of pronominals 11.1.4 Genitive 10.3.8
- of pronouns 11.1.4 - Autonomous 10.3.8 [A]
- of verbals 9.1 - Choice 10.3.8 [C]
Functional equivalence 6.2.2 - Classifying 10.3.8 [A], 12.2.5 [C]
Functional Grammar 1.1,6.3.2 - Local 10.3.8 [A]
Functional stability 5.2 - Objective 10.3.8 [B]
Functions of the passive 7.4.2 - Post-10.3.8 [A]
Future forms - Specifying 10.3.8 [A]
- Deontic use 9.7 - Subjective 10.3.8 [B]
- Epistemic use 9.7 - Syntactic function 10.3.8 [A]
- Past future 9.6.3 - and partitive construction 10.3.8 [A]
- Present future 9.6.3 Genitive case 11.2.1 [D]
Future of present cause 9.6.3 Genitive construction 10.3.8
Future perfect forms 9.6.5 Genitive form 8.3
- Past future perfect 9.6.5 Genitive marker 10.3.8
- Present future perfect 9.6.5 German 5.3.5, 8.10 [B], 10.2.3,12.1.7
Future perfect progressive forms GET-passive 7.4.5
- Past future perfect progressive Given information 5.2.3, 7.4.2
9.6.10 Global negation 7.5.3, 8.11
- Present future perfect progressive Government 6.3.3
9.6.10 Gradable adjective 12.1.2
Future progressive forms Grammatical function words 3.1.4 [C]
- Past future progressive 9.6.8 Grammatical functions 4.4.2
- Present future progressive 9.6.8 Grammatical modality 9.9.1
Futurity 9.6.3 Grammatical relations 5.2.1
Grammatical structure 2.3-5
g-replacive one 11.3.4 [E] Grammatical variation 1.6
Gender Grammatical verb 9.5
- and indefinite pronouns 11.3.4 [A.b] Grammaticalization 2.1,9.6.3
512 Subject index
Greek 5.1,10.4.2,10.3.8 - Determiner function 10.3.1

Group 3.1.3, 3.3.1,4.4.3 Indefinite class-member reference
- discontinuous 3.2.2 10.3.7 [A]
Group genitive 10.3.8 Indefinite determination 10.1.4,10.3.1
Indefinite expression 8.4
Habit (stative situation) 7.2.3 Indefinite mass reference 10.3.7 [B]
Habitual predictability 9.7 Indefinite non-specific reference 10.3.6
Habitual situation 9.6.2,9.6.7 [C]
Head 3.3.1 Indefinite pronoun 11.3.4; 11.1.2
Hebrew 5.1 - Autonomous use 11.3.4 [A.a]
Hierarchic 2.5 - Determinative use 11.3.4 [A.a]
Highlighting 5.2.3, 5.6.1,8.4 - and assertiveness 11.3.4 [A.fJ
Historic present 9.6.2 - and case 11.3.4 [A.c]
Historical narrative - and determination 11.3.4 [A.a]
- and tense-aspect forms 9.6.14 - and gender 11.3.4 [A.b]
HOLDER (Specific participant role) - and non-assertiveness 11.3.4 [A.f]
7.32-4 - and number 11.3.4 [A.d]
Homonymy 3.1.4 - and ^-construction 11.3.4 [A.a],
Hypotaxis 12.1.5, 12.2.4 11.3.4[A.g]
Hypothetical meaning 8.9,9.7, 9.8.3 - and partitive construction 11.3.4
[A.a]
Identification 8.4 - and partitive meaning 11.3.4 [A.f]
Indefinite article - and post-modification 11.3.4 [A.a]
- and quantification 10.3.6 [H] - and pre-modification 11.3.4 [A.a]
Indefinite pronoun group - and reference 11.3.4 [A.g]
- and adjectives 12.1.7 - and syntax 11.3.4 [A.a]
i/-clause 8.9 - and universal meaning 11.3.4 [A.fj
Illocutionary converter 8.10 [B] Indefinite pronoun group 12.1
Illocutionary dimension (interacting) Indefinite pronouns 3.1.4,7.6.3 [D]
4.2.2 Indefinite quantified subjects 7.5.8
Illocutionary value 4.4.2, 5.2.2, 5.6.2 Indefinite specific reference 10.3.6 [C]
- and partial inversion 5.3.3 Independent relative pronoun
Imperative 9.8.1,9.8.2 - Determiner function 10.3.1
- Deontic use 9.8.1 - and reference 11.3.3 [B.c]
- Emphatic 9.8.2 Independent relative clause 5.4.1,7.6.3
- Prohibitive 9.8.2 [C], 8.2-4, 8.6, 11.3.3 [B.c]
- and voice 9.8.2 Indian English 1.6
- with LET 9.8.2 Indicative mood 9.8.1
Imperative clause 4.4.4 Indirect directive 8.1
Imperative conditional sentence 8.9 Indirect object 3.2.6; 3.2.4,4.3.1-3
Imperative construction 4.4.2-3, 9.2 - clause 8.7
Imperative sentence 4.4.2-4, 5.2.2, Indirect speech
7.5.9 - and past perfect 9.6.4
Implicit indirect speech - and past future progressive 9.6.8
- and tense-aspect forms 9.6.13 — and tense-aspect forms 9.6.13
In medias res perspective 9.6.1, 9.6.6 Indirectly reported speech 8.5
Inclusive present 9.6.2 Inferential conditional sentence 8.9
Indefinite article 10.3.6 Infinitive 3.1.4, 3.2.1
Subject index 513
- Accusative with ... 5.6.1 - and case 11.3.2 [B.a]

- Nominative with ... 5.6.1, 7.4.3 [A], - and gender 11.3.2 [B.b]
7.4.4 [A], 7.6.3 [C] - and syntax 11.3.2 [A]
- Split 5.5.6 Interrogative scope
Infinitive clause 3.3.2, 8.2 - Qualitative selectivity 11.3.2 [B.c]
Infinitive construction 9.2 - Quantitative selectivity 11.3.2 [B.c]
Infinitive marker to 3.1.4 Interrogative sentence 4.4.2-4, 5.2.2,
Infinitive verb group 9.4 7.5.9
Inflectional co-variation'7.6.2 Interrogative subclause 11.3.2 [A]
Inflections 1.2,2.1,5.6.5 Intonation 1.7,2.2,4.4.1,4.4.2
Informal style 1.4-7 Intransitive P
Information structure 5.2.3 - and predicative adjective 12.1.4
- and full inversion 5.3.5 Intransitivity 3.2.3-5, 3.2.7,4.3.1, 5.3.6
-ing form 9.2 Intrasentential ellipsis 4.2.2
-ing participle clause 3.3.2, 8.2-3, 8.6-8 Introspection 1.3
Inherent adjective 12.1.2 Inversion 5.3; 5.1, 5.2.2,6.4.5, 7.5.5
Inherent Mod. adjectives 12.2.3 - Full subject-predicator ... 5.3.2,
Initial position of adverbials 5.5.2 11.2.1 [E]
Instantaneous present 9.6.2 - Partial subject-predicator ... 5.3.2,
INSTRUMENT (Specific participant 8.9, 11.3.2 [A]
role) 7.3.2^ Inversion of sentence constituents
Intensive plural 10.4.3 5.3.10
Intensive relations (stative situation) Irish English 1.6
7.2.3 Irregular comparison
Interdependent relation 6.3.2 - of adjective 12.3.3
Interjection 3.1.4,4.5 - of adverb 12.3.3,12.5.5
Internal concord 7.6 Irregular plural 10.4.2
Internal discontinuity 5.6.3, 5.6.5 Irregular verb 9.2
Internal focus Irreversible coordination 6.2.3
- and progressive forms 9.6.6 Isomorphism 6.2.6
Internal order it
- of adjectival mod. I zone 12.2.5 [A] - uses 11.2.1 [G]
- of adjectival mod. II zone 12.2.5 [B] Italian 5.1
- of adjectival mod. Ill zone 12.2.5 Iterative (dynamic) situation 7.2.2
[C] - and progressive form 9.6.6
Internal order
- of adjectival mod. zones 12.2.5 Latin 5.1, 10.2.3,10.4.2
Internal structure Left dislocation 4.5
- of noun group 10.1.3 Left-hyphenation 3.1.6
- of verb group 9.4 Legal English 1.4
Interrogative clause 4.4.4, 5.4.2, 5.6.1, Lexical compounding 6.3.3, 12.1.3
8.2 Lexical item 3.1.4 [B]
Interrogative conditional sentence 8.9 Lexical meaning 3.2.1, 9.5
Interrogative determination 10.3.1 Lexical modality 9.9.1
Interrogative pronoun 3.1.4, 6.4.2, Lexical word 10.1.4
7.6.3 [D], 11.3.2; 11.1.2 Lexicon 2.1
- Autonomous use 11.3.2 [A] LGP (Language for General Purposes)
- Determinative use 10.3.1,11.3.2 [A] 1.8
514 Subject index
Linear analysis 3.1.2 - Lexical 9.9.1

Linearity (Principle of...) 2.4 - and desirability 9.7
Linguistic ambiguity 2.6 - and non-factuality 9.8.1
Linguistic competence 2.7 - and probability 9.7
Linguistic context 4.2.2, 7.4.3 - expressed by collocations 9.9.1
Linguistic creativity 2.6 - in hypothetical-conditional
Linked coordination 3.3.3,6.2.3, 12.1.5 sentences 9.9.2-3,9.9.5-6
Linking 6.1 - in indirect speech 9.9.2-3,9.9.6
- Causal 8.9 - in interrogative sentences 9.9.2-5
- Temporal 8.9 - in negative sentences 9.9.2-6
Literary narration 12.1.7 - in sentences in the past 9.9.2-4
- and tense-aspect forms 9.6.14 - in yes-no interrogative sentences
Local genitive 10.3.8 [A] 9.9.4
Local negation 7.5.6; 7.5.3, 8.11 Modification 6.3.3 [C], 10.1.4,11.1.4,
LSP (Language for Special Purposes) 12.1.1
1.8 - Adjectival 12.2
-fy adverb 12.5.2 - Descriptive 10.1.4
- Specifying 10.1.4
Main clauses 3.3.4, 8.9 Modification
Maledictions 5.3.3,9.9.2 - of adverbals 12.5.3
Mandative (deontic) subjunctive 9.8.3 Modificational zones 12.2.2-3
Mass noun 10.2.5,10.3.5 [A], 10.3.6 - Internal order 12.2.5
[F], 10.3.7 [B], 10.3.7 [E] - Structure 12.2.4
Matrix clause 3.3.4, 5.6.1, 6.4.1, 9.6.3, Modifier
9.6.7 - Noun group as ... 10.4.6
Maximal expansion strategy 6.2.1 Monocentric view of English 1.5
Meaning 2.1 Monotransitivity 3.2.4, 3.2.6,4.3.2
Medial ellipsis 4.2.2 Mood 9.8
Medial position of adverbiale 5.5.3 - Imperative 9.8.1
Medium 1.7; 1.4,1.8 - Indicative 9.8.1
Mesolect 1.6 - Subjunctive 9.8.1
Middle English 1.5 Morphemes 2.1-2
Middle verbs 7.4.5 Morphological comparison 5.6.5
Missing constituent 4.2.1-2 - Choice 12.3.4
Missing forms 6.2.1 - of adjective 12.3.1
Mixed coordination 6.2.2 - of adverb 12.5.5
Mobility 5.1 Morphology 2.1
Modal auxiliaries 1.6, 1.8, 3.2.1, 3.3.1, - and adverb 12.5.2
4.1.2,7.5.8,7.6.2,9.5,9.9.1 - of adjectives 12.1.3
- Order 9.4 - of verbs 9.2
- and fmiteness 9.4 Movement 5.1
Modal use MUCH vs. VERY 12.5.3
- of tense-aspect forms 9.7,9.9.1 Multi-functional zone 10.1.4
Modal verb 1.6, 3.2.1,9.5 Multiple coordination 6.1
Modality 9.9 Multiple subordination 6.1
- Deontic 9.7 Multipliers 10.3.3
- Epistemic 9.7
- Grammatical 9.9.1 n-replacive one 11.3.4 [E]
Subject index 515
Name 10.2.4 - and indefinite pronouns 11.3.4 [A.fJ

- and countability 10.2.5 Non-countable quantification 10.1.4
- and proper noun 10.3.5 [D] Non-declarative clause 4.4.4
Narrative 9.6.7 Non-declarative sentence 7.5.9
- Literary 12.1.7 Non-deictic use
Narrative continuity 1.7, 5.3.6 - of demonstrative pronoun 11.3.1 [E]
Narrative Non-factuality 9.7,9.8.1,9.9.1
- and past verb forms 9.6.14 Nonfinite clause 3.3.2, 3.3.4, 5.4.4, 8.2-
- and tense-aspect forms 9.6.14 3, 8.5-6, 8.8
Native speakers] 2.6-8; 1.1,1.3,4.4.1 Nonfinite passive 7.4.5
Near pronoun 11.3.1 [D] Nonfinite predicator 9.1
Necessity 9.9.4 Nonfinite progressive forms 9.6.11
Negation 7.5 Nonfinite subclause 11.2.3 [A]
- Clausal 7.5.4, 7.5.6 - and reflexive pronoun 11.2.3 [A]
- Domain 7.5.3 Nonfinite verbs 6.4.5
- Double 7.5.5 Nonfiniteness 3.2.1, 3.3.4
- Global 7.5.3-8,8.11 Non-gradable adjective 12.1.2
- Implicit 7.5.7 Non-inherent adjective 12.1.2
- Limited 7.5.4-6 - and subject complement 12.1.2
- Local 7.5.6; 7.5.3, 8.11 Non-recursive coordination 6.2.3
- Quantifier... 7.5.5 Non-referential grammatical prop word
- Semantic scope 7.5.8 3.2.2,3.2.5, 11.2.1 [G]
- Syntacticfield7.5.4-5, 7.5.8 Non-restrictive adjective 12.1.2
- and adverbials 7.5.8 Non-restrictive contrast-formation
- and modal auxiliaries 7.5.8 10.1.4
Negation transfer 8.11 Non-restrictive relative clause 5.4.1,
Negative adverb 7.5.5 11.3.3 [A]
Negative concord 7.5.5 - reduced 8.2
Negative expression 8.4 Non-textual ellipsis 4.2.2
Negative opener 5.3.4 Non-textual reference 10.3.4
Negative prefix 7.5.6 Notional concord 7.6.4
Negative pronouns 7.5.5 Notional passive 7.4.5
Negative quantifiers 7.5.5 NOT-contraction 7.5.2
Negative sentences 5.6.2, 7.5.1 Noun 3.1.4
New information 5.2.3, 7.4.2 - Aggregate 10.4.4
New Zealand English 1.6 - Collective 7.6.4 [A], 10.4.4
NICE-properties 9.5 - Collectivizing 10.4.6
Nominal clauses 3.2.7-8, 8.2 - Common 10.2.4
Nominal constructions 3.2.7-8 - Compounds and plural 10.4.5
Nominals3.3.1,10 - Count 10.2.5,10.3.5 [A], 10.3.6 [E],
- Functional domain 10.1.4 10.3.7 [A], 10.3.7 [E]
- Functions 10.1.2 - Gender 10.2.3
- and reference 10.1.4 - Mass 10.2.5,10.3.5 [A], 10.3.6 [F],
Nominative case 11.2.1 [D] 10.3.7 [B], 10.3.7 [E]
Nominative with infinitive 5.6.1, 7.4.3 - Number-invariable 7.6.3 [A] 10.4.3
[A], 7.4.4 [A], 7.6.3 [C] - Plural 10.4
Nonassertive forms 7.5.7; 7.4.4 [B] - Plural used as singular 10.4.3
Nonassertiveness - Proper 10.2.4,10.3.5 [A], 10.3.6 [G]
516 Subject index
- Semantics 10.2.2 OPT (optative) 4.4.2

- Singular of 10.4 Optional adverbials 3.2.9, 7.3.4, 8.1
- Singular used as plural 10.4.3 Optional non-characterizing element
- Type 10.2.4 3.3.1
- and deixis 10.3.5 [D] Order of auxiliaries 9.4
- in plural 10.2.5 Ordinal numbers 10.3.3
- in singular 10.2.5 Ordinals 10.3.3,11.3.4 [E]
Noun clause 3.2.7-8
Noun group 3.3.1 Paradigmatic relationship 2.8, 3.1.1
- Discontinuous 5.6.3 Paralinguistic means 1.7
- Internal structure 10.1.3 Parataxis 12.1.5
- and post-head dependent 10.1.3 Partial inversion 3.2.1,5.3.2, 8.9,
- and pre-head dependent 10.1.3 11.3.2 [A]
- as modifier 10.4.6 - and initial constituent 5.3.4
Noun group tag 4.5 - with illocutionary value 5.3.3
Number 7.6.2 Participant roles 7.3, 8.2
- and demonstrative pronoun 11.3.1 - Double role analysis 7.3.4
[A] Participial verb group 9.4
- and indefinite pronouns 11.3.4 [A.d] Participle construction 9.2
Number and weight Participles 3.2.1
- and plural 10.4.6 - Adjectival 12.1.3
Number-invariable noun 7.6.3 [A], - Dangling 8.8
10.4.3 - Past 3.2.1,3.2.5, 5.3.7
Number-transparent expressions 10.2.1 - Present 3.2.1,5.3.7
Numerals 3.1.4 - Unattached 8.8
Partitive construction
Object 3.2.6, 3.2.3-4 - and genitive 10.3.8 [A]
- Provisional 3.2.5, 11.2.1 [G] - and indefinite pronouns 11.3.4 [A.a]
- and choice of relative pronoun Partitive expression 10.2.5
11.3.3[C.d] Partitive meaning
Object clause 8.3, 8.5 - and indefinite pronouns 11.3.4 [A.fJ
- Discontinuous 5.6.1 Passive 3.2.5-6, 3.2.8,4.3.2-3, 7.4
Object complement 3.2.8; 3.2.4 - Constraints 7.4.4 [B]
- clause 8.7 - Extended use 7.4.3
Objective case 3.2.5 - Functions 7.4.2
Objective genitive 10.3.8 [B] - Nonfinite 7.4.5
Obligatory adverbial 3.2.9 - Notional 7.4.5
Obligatory characterizing element 3.3.1 - Prepositional meaning 7.4.4 [B]
Obligatory non-characterizing element - and reflexive pronouns 11.2.3 [A]
3.3.1 - and stranded prepositions 7.4.3 [B]
(/-construction 5.6.5,12.3.5,12.4.2 - voice and stative situations 7.4.3 [A]
- and indefinite pronouns 11.3.4 [A.a], - without active 7.4.4 [A]
11.3.4[A.g] Past [verb form] 8.9
- and possessive pronouns 11.2.2 [A] Past 9.6.2
- and specifying genitive 10.3.8 [C] Past future [verb form] 8.9
Open condition 8.9 Past future perfect [verb form] 8.9
Open word class 3.1.4 Past future progressive 9.6.8
Operator 3.2.1-2,4.1.2 - and indirect speech 9.6.8
Subject index 517
Past participle 3.2.1, 3.2.5, 5.3.7 Pluricentric view of English 1.5

- Adjectival 12.1.3 Polarity 7.5, 8.11
Past participle clause 3.3.2 - Reversed 8.10 [B]
Past perfect progressive 9.6.9 Polarity in complex sentences 8.11
Past perfect 8.9, 9.6.4 Positional restrictions
- and backshifting 9.6.4 - of adjectives 12.1.7
- in indirect speech 9.6.4 Positive 5.6.5,12.3.1
PER (Performative) 4.4.2 Possessive pronoun 3.1.4,10.3.8,
Perceptions (stative situation) 7.2.3, 11.2.2,11.1.2
7.4.3 [A] - Autonomous use 11.2.2
Perfect forms - Determinative use 11.2.2
- Nonfinite 9.6.4 - Determiner function 10.3.1
- Past perfect 9.6.4 - and definite article 11.2.2 [C]
- Present perfect 9.6.4 - and o/ônstruction 11.2.2 [A]
Perfect progressive forms - and OWN 11.2.2[D]
- Past perfect progressive 9.6.9 - and personal pronouns 11.2.2 [D]
- Present perfect progressive 9.6.9 - and post-genitives 11.2.2 [B]
Performance 2.7 Possibility 9.9.2,9.9.3
Performative 4.4.2 Post-genitive 10.3.8 [A]
Peripheral adjective 12.1.6 - and possessive pronouns 11.2.2 [B]
Permanentness Post-head dependent 10.1.3, 12.1.1
- and adj ective 12.1.2 - in noun group 10.1.3
Permission 9.9.2,9.9.3 Postmodification 6.3.3
Person 7.6.2 - and indefinite pronouns 11.3.4 [A.a]
Personal pronoun 3.1.4,11.2.1; 11.1.2 Postcedent 10.3.4
- and case 11.2.1 [D] Postdeterminers 10.3.2
- and gender 11.2.1 [C] Postponement 5.2.3, 5.6.1
- and number 11.2.1 [A] Pragmatic function 3.2.5
- and person 11.2.1 [B] Pragmatics 1.8
- and possessive pronoun 11.2.2 [D] Pre-head dependent 10.1.3
- and reference 11.2.1 [F] - in noun group 10.1.3
- and syntax 11.2.1 [E] Predeterminers 10.3.2
Phonotactic rules 9.2 Predicate 4.1.2,6.2.2
Phrasal verb 4.3.1-3, 11.2.1 [E] Predicate position
Phrasal-prepositional verb 4.3.3 - of adjective 12.1.7
PLACE (Specific participant role) Predicate stack 4.1.2,4.2.3
7.3.2-4 Predication 4.1.2, 6.2.2
Plural Predication stack 4.1.2
- Distributive 10.4.6 - and full inversion 5.3.7, 5.6.2
— Intensive ... of noun 10.4.3 Predicative adjective 12.1.4
- Irregular 10.4.2 — and intransitive predicator 12.1.4
- fact expressions 7.6.4 [D] Predicative use
- form as singular 7.6.3 [A] - of adjective 12.1.4
- of compounds 10.4.5 Predicator 3.2.1,7.6.2
- of nouns 10.4 - Copula 3.2.3
- of number and weight 10.4.6 - Finite 3.2.1,9.1
Plural noun 10.2.5 - Intransitive 3.2.3
Plural number 7.6.3 - Transitive 3.2.3
518 Subject index
Predictability - and self-contained situations 9.6.6

- General 9.7 - and telic situations 9.6.6
- Habitual 9.7 - and temporalframe9.6.7
- Nonfuture9.7 - and actionality 9.6.6
- Specific 9.7 - and external focus 9.6.6
Premodification 6.3.3 - and internal focus 9.6.6
- and indefinite pronouns 11.3.4 [A.a] Pronominal 1 1
Preposition group 3.3.1-2; 4.3.2,6.3.3 - External relations 11.1.3
[D] - Functional domain 1 1 . 1 .4
- Discontinuous 5.6.4 Pronoun 11; 3.1.4, 4.2.3, 6.3.3 [A]
- and reflexive pronoun 11.2.3 [A] - Classification 1 1 . 1 .2
Preposition 3.1.4 - Demonstrative 10.3.1
- Stranded 3.3.2,4.3.2,7.4.3 [B] - Functional domain 1 1 . 1 .4
— and choice of relative pronoun - Indefinite 7.6.3 [D], 11.3.4
11.3.3[C.a],[C.d] - Interrogative 6.4.2, 7.6.3 [D], 10.3.1,
- vs. adverb 12.5.2 11.3.2
Prepositional complement 3.3.2 - Personal 7.6.3 [D], 11.2.1
Prepositional object 4.3.2 - Possessive 10.3.1, 10.3.8
Prepositional predicator analysis 4.3.2 - Reflexive 11. 2.3
Prepositional verb 3.3.2,4.3.2 - Relative 10.3.1
Present (verb form) 3.2.1-2,6.2.4, - Representative use 1 1 . 1 .4
7.6.2, 8.9,9.6.2 - Specifying use 1 1 . 1 .4
Present future 9.6.3 - Syntactic use 11.1.3
Present future progressive 9.6.8 - Universal 11. 3.4 [B-C]
- Backshifting 9.6.8 - and reference 1 1 . 1 .4
Present indicative construction 9.2 - and referent 11.1.4
Present participle 3.2.1, 5.3.7 Pronoun group 3.3.1,11.1.1
Present perfect 1.6,7.4.4 [A], 9.6.4 - Definition 11. 1.1
Present perfect progressive 9.6.9 - Discontinuous 5.6.3
Presupposition 8.4 Pronunciation 9.2, 12.3.2
Primary verb 9.5 Prop word 3.2.2, 3.2.5, 1 1.2.1 [G]
Primary verbs 3.2.1, 5.3.2 Proper noun 7.6.4 [C], 10.2.4, 10.3.5
Principle of end-weight 5.2.3 [A], 10.3.6 [G]
Principle of increasing complexity - and name 10.3.5 [D]
5.2.3 Prosody 1.7, 10.2.1, 12.3.2
Principle of linear modification 5.5.10 Protasis 8.9
Principle of linearity 2.4; 2.5, 2.8, 2.9 Provisional object 3.2.5, 11.2.1 [G]
Principle of proximity 2.4, 5.2.1 Provisional subject 3.2.2, 5.3.7, 7.6.3
Probability
- and modality 9.7 Proximity (Principle of . . .) 2.4
Pro-forms 4.2.3 Pseudo-cleft sentence 8.4
Progressive construction 9.2 Pseudo-coordination 6.2.3
- Nonfinite9.6.11 Punctual (dynamic) situation 7.2.2
- Past 9.6.7 - and progressive form 9.6.6
- Present 9.6.7 Putative meaning 9.9.5
- and directed situations 9.6.6
— and iterative situations 9.6.6 Qualification (modal) of utterance 9.9.1
- and punctual situations 9.6.6 Qualitative selectivity 11.3.2 [B.c]
Subject index 519
Quantification 10.4; 6.3.3 [D], 10.1.4, - and indefinite pronouns 11.3.4 [A.g]
11.1.4 - and independent relatives 11.3.3
- Countable 10.1.4 [B.c]
- Non-countable 10.1.4 - and personal pronoun 11.2.1 [F]
- and determination 10.3.3 - and pronoun 11.1.4
- and indefinite article 10.3.6 [H] - and relative pronouns 11.3.3 [B.c]
- of count noun 10.2.5 - in nominals 10.1.4
- of mass noun 10.2.5 - to persons with demonstrative pro-
Quantifier 6.3.3 [D], 11.1.2,11.3.4 [E] nouns 11.3.1 [B]
- Negative 7.5.5 Referent 7.2.1,10.2.4
Quantitative selectivity 11.3.2 [B.c] - Class-member 10.2.4
Quantity expressions 7.6.4 [B] - Generic 10.2.4
QUE [Question] 4.4.2 - Type 10.2.4
Question 5.3.3 - Unique 10.2.4
- and pronoun 11.1.5
Raising of negation 8.1 Referential orientation 10.3.4
Raising of subject 5.6.1 Reflexive pronoun 11.2.3; 3.1.4, 5.6.3,
Rank and adjectives 12.3.5 11.1.2
Real object 3.2.5 - Emphatic dependent use 11.2.3 [B-
Real subject 3.2.2 C]
Reciprocal verb 7.4.4 [B] - Reflexive use 11.2.3 [A]
Recursive coordination 6.1,6.2.3 - and nonfinite subclause 11.2.3 [A]
Recursiveness 3.3.5 - and passivization 11.2.3 [A]
Reduced cleft sentence 8.4 - and preposition groups 11.2.3 [A]
Reference 7.2.1, 10.1.4, 10.2.4,10.3.4 Reflexive use
- Anaphoric 8.10 [C], 10.3.4,11.2.1 - of reflexive pronoun 11.2.3 [A]
[F] Regular singular/plural 10.4.1
- Cataphoric 8.10 [C], 10.3.4,11.2.1 Regular verb 9.2
[F] Relative adjunct 11.3.3 [B.c], 11.3.3
- Class-member 10.3.6 [A]; [C.e]
- Definite non-specific 10.3.5 [B] Relative adverb 6.4.2
- Definite specific 10.3.5 [A] Relative clauses 5.4.1
- Endophoric 10.3.4 - Independent 5.4.1, 7.6.3 [C], 8.4,
- Exophoric 10.3.4 11.3.3 [B.c]
- Generic 10.3.5 [C], 10.3.7 [C], - Non-restrictive 5.4.1, 8.4, 11.3.3 [A]
12.4.2 - Restrictive 8.4, 11.3.3 [A]
- Indefinite class-member 10.3.7 [A] - Sentential 8.10 [C]
- Indefinite mass 10.3.7 [B] Relative determination 10.3.1
- Indefinite specific 10.3.6 [C] Relative pronoun 11.3.3; 3.1.4,6.4.2,
- Indefinite non-specific 10.3.6 [C] 11.1.2
- Non-textual 10.3.4 - Autonomous use 11.3.3 [A]
- Specific 12.4.2 - Choice 11.3.3 [C]
- Sub-generic 10.2.5 - Determinative use 11.3.3 [A]
- Textual 10.3.4 - Determiner function 10.3.1
- Type 10.3.4 - Independent 10.3.1,11.3.3 [B.c]
- Unique 10.3.5 [D] - and anaphoric reference 11.3.3 [B.c]
- and demonstrative pronoun 11.3.1 - and case 11.3.3 [B.b]
[D] - and cataphoric reference 11.3.3 [B.c]
520 Subject index
- and gender 11.3.3 [B.a] - of adverbal 3.2.9, 7.3.2, 8.8,12.5.1

- and syntax 11.3.3 [A] - of noun 10.2.2
Relati vization 3.3.2 - of specifying genitive 10.3.8 [B]
Report 9.9.5 Semi-auxiliary 9.5, 9.9.1
Reported speech 8.5 Sent 3.1.2,4.4.1,4.4.3
Reporting clause 5.3.9 Sentence 2.2
Representative use of pronoun 11.1.4 - Conditional 9.6.3
Restrictive adjective 12.1.2 - Exclamatory conditional 8.9
Restrictive contrast formation 10.1.4 - Imperative conditional 8.9
Restrictive relative clause 8.4, 11.3.3 - Inferential conditional 8.9
[A] - Interrogative conditional 8.9
RESULT (Specific participant role) Sentence constituents and inversion
7.3.2-4 5.3.10
Retrievability 6.2.1 Sentence functions 3.1.2
Reversed polarity 8.10 [B] Sentential relative clause 8.10 [C]
Reversible coordination 6.2.3 Simple adverb 12.5.2
Rheme 5.2.3 Simple coordination 6.2.3
Right-dislocation 4.5 Simple past 1.6,9.6.2
Role suppression 7.3.4 Simple predicator analysis 4.3.2
Russian 5.1 Simple sentence 7.1
Singular determiners 7.6.4 [A]
-j form 9.2 Singular form as plural 7.6.3 [A]
- Scandinavian 5.3.5 Singular number 7.6.3
Scope 5.5.10, 5.5.13,9.6.3 Singular
Scottish English 1.6 - of nouns 10.2.5,10.4
Second Language 1.4 Singular/plural distinction 10.4.1
Self-contained (dynamic) situation Situation types 7.2.1
7.2.2 Situational context 4.2.2, 7.4.3
- and progressive form 9.6.6 Situational focus 9.6.1
Self-reference avoidance 7.4.2 Situations 7.2, 9.1
Semantic function of adverbial clause - Dynamic 7.2.1,9.6.4
8.8 - Stative 7.2.1,9.6.4
Semantic functions of constituents Slot 2.8, 3.1.1
3.2.5 South African English 1.6
Semantic scope of negation 7.5.8 Spanish 7.5.5
Semantic weight 3.2.1 Speaker-oriented adverbial 3.2.9
Semantics Specific participant roles 7.3.2
- Dynamic adjective 12.1.2 Specific predictability 9.7
- Gradable adjective 12.1.2 Specific reference 12.4.2
- Inherent adjective 12.1.2 Specification
- Non-gradable adjective 12.1.2 - Adjectival subfunction 12.2.2
- Non-inherent adjective 12.1.2 SPECIFIED (General participant role)
- Non-restrictive adjective 12.1.2 7.3.1-4,10.3.8 [B]
- Permanentness 12.1.2 SPECIFIER (General participant role)
- Restrictive adj ective 12.1.2 7.3.1-4
- Stative adjective 12.1.2 Specifying adjective 12.1.6
- Temporariness 12.1.2 Specifying genitive 10.3.8 [A]
- of adjectival 12.1.2 - Semantics 10.3.8 [B]
Subject index 521
- and »/"-construction 10.3.8 [C] Subject retainment 7.4.2

Specifying modification 10.1.4 Subjective case 3.2.5
Specifying use of pronoun 11.1.4 Subjective genitive 10.3.8 [B]
Speech act 4.4.1,4.4.3, 5.2.2 Subjectless clause 8.2
Split infinitive 5.5.6 Subject-operator inversion 3.2.1
Spoken English 1.7 Subject-predicator concord 3.2.2, 7.6,
Spoken vs. written 1.4 7.8
STA [Statement] 4.4.2 Subject-predicator inversion 5.3
Stacking 4.1.1-3; 4.3.2,4.4.4,10.1.1-2 - Full 5.3.2
Standard American English 1.5 - Partial 5.3.2
Standard British English 1.5 Subjunctive 1.7,4.4.2,9.8.3
Standard English 1.5 - Concessive (deontic) 9.8.3
Standard negation 7.5.2,7.5.5 - Mandative (deontic) 9.8.3
State 7.2.3 - and independent relative clauses
Statement 4.4.2 9.8.3
Static vs. dynamic 1.7 - and yes-no interrogatives 9.8.3
Stative adjective 12.1.2 Subjunctive construction 9.2
Stative situations 7.2.1,7.4.3 [A], 9.6.2 Subjunctive mood 9.8.1
- Attitudes 7.2.3 - Deontic use 9.8.1,9.8.3
- Extensive relations 7.2.3 - Epistemic use 9.8.1,9.8.3
- Habit 7.2.3 Subordinate clauses 3.3.4, 5.4, 8.1-2
- Intensive relations 7.2.3 - Syntactic function 8.2
- Perceptions 7.2.3 - and constituent order 5.4
Stative verbs 3.2.7-8 Subordinating conjunctions 3.3.4,
Stative/dynamic distinction 3.2.7-8, 5.3.4,6.4.1
7.3.2 - Complex 6.4.3
Stranded prepositions 4.3.2, 7.4.3 [B] Subordination 1.7, 3.3.4-5,4.1.1,4.2.1,
Stress 1.7,2.3 6.1,6.3,6.4.2,6.4.5
Structure - Binary 6.1
- of adjectival mod. zones 12.2.4 - Multiple 6.1
Subclauses, see Subordinate clauses - Prosodically signalled 6.4.5
Sub-generic reference 10.2.5,10.3.7 - Recursive 6.1
[C] - at clause level 6.3.2
Subject 3.1.2, 3.2.1-2,7.6.2 - at group level 6.3.3
- Provisional 5.3.7, 7.6.3 [D], 11.2.1 Subordinators 8.9
[G] - Correlative 6.4.4
- and choice of relative pronoun Substantival use
11.3.3[C.c] - of adjective 12.4
- complement concord 7.7 Superlative adjective 11.3.4 [E]
Subject clauses 7.6.3 [C], 8.3 Superlative degree 5.6.5, 12.3.1,12.3.5
- Discontinuous 11.2.1 [D] Supplementive clause 8.8
Subject complement 3.2.4,3.2.7,11.2.1 Syntactic comparison
[D] - of adjective 5.6.5, 12.3.1
- clauses 8.6 - of adverb 5.6.5, 12.5.5
- and non-inherent adjectives 12.1.2 Syntactic comparison of adjective
Subject position 3.2.2 [A], 11.2.1 [D] - Choice 12.3.4
Subject raising 5.6.1 Syntactic discord 7.6.4
Subject realization 7.6.3 Syntactic field of negation 7.5.4
522 Subject index
Syntactic function - Past 9.6.1

- of genitive 10.3.8 [A] - Past future 9.6.1
Syntactic patterns 3.2.3, 3.2.4 - Past future perfect 9.6.5,9.6.1
Syntactic property of cleft sentence 8.4 - Past future perfect progressive 9.6.1,
Syntactic units 2.1 9.6.10
Syntactic use - Past future progressive 9.6.1,
- of pronoun 11.1.3 9.6.8
Syntactic zones - Past perfect 9.6.1
- Mod. 112.2.2 - Past perfect progressive 9.6.1,9.6.7,
- Mod. II 12.2.2 9.6.9
- Mod. Ill 12.2.2 - Past progressive 9.6.1,9.6.7
Syntagmatic relationship 2.8 - Present 9.6.1
Syntax 2.5 - Present future 9.6.1
- Domain 2.2 - Present future perfect 9.6.1,9.6.5
- and indefinite pronouns 11.3.4 [A.a] - Present future perfect progressive
- and interrogative pronouns 11.3.2 9.6.1,9.6.10
[A] - Present future progressive 9.6.1,
— and personal pronoun 11.2.1 [E] 9.6.8
- and relative pronouns 11.3.3 [A] - Present perfect 9.6.1
Systemic Grammar 1.1 - Present perfect progressive 9.6.1,
9.6.9
Tag clause 8.10 [B] - Present progressive 9.6.1,9.6.7
Tag question 3.2.2,4.5, 5.3.3, 7.5.2, Tentativenes 9.9.2,9.9.3
8.10[B] Terminal ellipsis 4.2.2
Telic [dynamic] situation 7.2.2 Terminal position of adverbiale 5.5.4
- and progressive form 9.6.6 Textual reference 10.3.4
Temporal frame 9.6.7 - Anaphoric 10.3.4
Temporal linking 8.9 - Cataphoric 10.3.4
Temporariness than -construction 5.6.5
- and adjective 12.1.2 fAaf-clause 8.2-3,8.6
Tense 9.6 - Elaborative 5.6.3,10.1.3
Tense-aspect Thematic meaning 7.4.2
- Functional-instructional approach Thematization 5.2.3,7.4.2,7.4.5
9.6.1 Theme 5.2.3
- in biographical narrative 9.6.14 Thought-mood 9.8.1
- in everyday narrative 9.6.14 TIME (Specific participant role) 7.3.2-4
- in historical narrative 9.6.14 Titles 7.6.4 [C]
- in literary narrative 9.6.14 Tone group 2.2,4.2.2,4.3.2
- relative distinctions 9.6.1, 9.6.6 Top of the tree 4.4
Tense-aspect forms Topic & comment 3.2.2,4.1.2
- Modal use 9.7,9.9.1 Topic (theme) 5.2.3
- andbackshifting9.6.13 Topicalization 5.2.3, 5.3.4, 5.6.1, 7.4.2
- and free indirect speech 9.6.13 to-infinitive clause 5.6.1, 5.6.3, 8.2-3,
- and implicit indirect speech 9.6.13 8.5-6, 8.8
- and indirect speech 9.6.13 Transfer of negation 8.11
- and literary narration 9.6.13 Transitive sentence 5.2.2
Tense-aspect system 9.6 Transitivity 3.2.3-7,4.3.1-4, 7.4.1
- Overview 9.6.12 Tree diagram 3.1.2
Subject index 523
Turkish 5.1 - Internal relations 9.4

Type Verb phrase 6.3.2
- of noun 10.2.4 Verbal 3.3.1, 9
- of reference 10.3.4 - Communicative 9.1
- of referent 10.2.4 - External relations 9.3
- Finite 9.1
Unattached participle 8.8 - Nonfmite9.1,9.3
Unique categorization 10.2.4 - complementation 6.3.2
Unique reference 10.3.5 [D] Verbless clauses 3.3.4, 5.4.4, 8.2,8.8
Unique referent 10.2.4 VERY vs. MUCH 12.5.3
Universal meaning Vocative 4.5
- and indefinite pronouns 11.3.4 [A.fJ Voice 3.2.5,3.2.8, 5.2.3, 7.4
Universal pronoun 11.3.4 [B-C] - Active 7.4.1
- Autonomous use 11.3.4 [B-C] - Passive 7.4.1
- Determinative use 11.3.4 [B-C] - Restrictions 7.4.4
Unlinked coordination 6.2.3 - and imperative mood 9.8.2
Unmarked determination 10.3.6 [A] Volition 9.9.6
Use of Volitional/non-volitional 7.3.2
— compared forms 12.3.5
Utterance 2.2 Weight and number
utterance pragmatics 9.9.4 - Use of plural 10.4.6
wA-interrogative 4.4.4, 5.2.2, 5.3.3,
Valency 7.3.1 5.4.2,7.5.9
Valency grammar 6.3.2 - clause 8.2-3, 8.6
Variety (formal/informal) 1.4,1.7 Will-mood 9.8.1
Variety (spoken/written) 1.4,1.7 Wishes 5.3.3,9.9.2
Variety (user-related/use-related) 1.4-8 - and deontic subjunctive 9.8.3
Verb 3.1.4,3.2.1, 3.3.1,4.3.1-4,9.1-9 Word (Single...) 3.1.3,4.4.3
- Base form 9.2 Word classes 3.1.4
— contraction 7.5.2 Word classes (closed vs. open) 3.1.4
ed form 9.2
[C]
- Grammatical verb 9.5
- -ing form 9.2
- Irregular verb 9.2 yes-no interrogative 3.2.1-2, 4.4.4,
- Morphologically defective 9.5 5.2.2,5.3.3, 5.4.2, 6.4.1, 7.5.9, 8.2-3,
- NICE-properties 9.5 8.6
- Primary 9.5
- -s form 9.2 Zero constituent 1.2,4.2.2-3
Verb group 3.2.1, 3.3.1, 4.3.1-4, 9.1 Zero convention 4.2.1-3
- Discontinuous 5.6.2 Zero determination 10.3.7
Word index
a/an 180,246, 360-1,390 analysis 393

aboard 491 ancient 449
about 86 and 166,172-6,178,213,241,
above all 442 248-9
abroad 452,490 (the) Andes 241,247
absent (v) 417 anger 241
acclaim 211 animate 450
according to 182 another 180,360-1,400,446
(the) accused 485 answer 264,344
acoustics 394 antelope 393
across 86,486 antenna 392
actor 354 antique 478
actress 354 any 165, 180, 228,360-1 363,400-1,
addendum 392 434,436-40,443,448
advise 254 (the) Americas 370
afloat 452,460 anybody 228,243,400,436
aforementioned 4 anyhow 260,437
aforesaid 4,11 anyone 228,260,400,436,440
afraid 448,450,452,478 anything 228,243,435,438,440
after 63,184,491 anyway 437
after all 442 anywhere 229,260,437
ain't 221 apart 490
airline 471 apart from 156
airs 395 apologies 395
alert 452,460 appear 43, 53, 153,276
alike 460 appendix 393
alive 460 apply 417
all 182,224, 244, 361-4, 390,400, aquarium 392
434-6,441,443,448 are 239,406
all in all 442 aren't 221
allow 153 as 272, 275, 372, 374, 408,412,418,
almost 493 448,459,487
(the) Almighty 485 as ... as 474
alone 452,460 as if 64, 188
aloof 460,478 as soon as 188
(the) Alps 241,247, 370, 394 as though 188
already 218,229 as well as 175,327
although 134,184 ashore 490
alumna 392 aside 490
alumnus 392 asleep 452,460,462,478
always 140,294 assume 195
am 221,239,406 assuming 119,270
among 427 at all 335,442
ample 476 athletics 394
526 Word index
(the) Atlantic 371 behave 417

Atlantis 352 believe 153,195,264,276,
atomic 449 298
attorney-general 396 belief 344
atypical 452 belong 298
audience 245,396 beneath 491
aunt 353 bestir 417
automatic 471 best 476
autumn 381 better 476
avail (v) 417 big 449,451,457,475,478
avoid 264 billiards 241, 394
awake 460 billy-goat 355
aware 460 binoculars 241,395
away 452 bizarre 478
black 463,466,471
bachelor 353 blatantly 487-8
backward(s) 488,490 blow 34
bacteria 392 board 396
bacterium 392 body politic 351,462
bad 449,451,457,476 boil 216
bake 216 book 341
band 245 boorish 451
banjo 391 boring 479
barely 226-7, 229 both 157, 361,363-4, 390,400,429,
barracks 393 436 443-4
basis 393 both ..'. and 170,176-7, 182,188,
be 39,42-3, 53, 117,214-5,217-8, 241-2,245
220-1, 235, 238-9, 259, 262, 272, bowl over 211
278,282-3,285-6,296,298, 311, boy 353
321-2,405 brace 393
be about to 95,292 brains 395
be bom 211 brave 450
be burned 217 bravo 391
be drowned 217 break 216
be going to 291-2 bride 354
be on the point of 95 bridegroom 354
be reputed to 211 bright 454
be rumoured to 211 brightly 454
be to 284,292 brilliant 465
bear (v) 214 broad 457
beat 278 brother 353,392
beautiful 449,451,470 buffalo 391,393
because 184,187 bureau 393
become 53, 218 burlesque 478
bed 381 bus 381, 390
bee 354 businesswise 490
before 63, 184,296-7 but 156,174-6,408,418,437
beg 254 by 174,297, 381,491
Word index 527
cactus 392 company 245

cancer 379 complete (adj) 450
can 39,220,274,283,285,325,327- concerned 460
32, 336 concerning 65
can't 334 (the) condemned 485
capable 449 conditions 380
car 381 conduct (v) 417
cards 394 congressional 449
careful 450 constitutional 451
carefully 487 contain 298
cargo 391 content (adj) 478
cattle 241,394 content (n) 395
certain 469 contents 241,395
chairperson 354 conviction 357
chance 344 cook (v) 216
change (v) 397 corps 393
charming 451 corpus 392
chief (adj) 469 could 269-70,274, 329-30
child 353, 392 council 396
Chinese 393,449,471,485 counsel 240,393
choose 432 countless 469
chorus 245 court martial 351,396,
church 381 462
civil 458,463 cowardly 488
claim (n) 344 crabwise 490
clamydia 379 crack (v) 216
class 245 craft 240, 393
clean (adj) 457,478 crash (v) 195
clergy 241,394 crew 394-5,420
clever 451,478 criminal 484
clockwise 486,490,494 crisis 393
close (v) 201,216 criterion 393
close (adv) 490 crowd 245
closely 490 cruel 475,479
cloud 358 cultural 471
(a) clove of garlic 357 custom 395
cock-pheasant 355 customs 241
coconut 357
cod 240 daily 490
codex 393 Danish 463
cold (adj) 457 dare 6, 284
college 381 dark 449
colours 241, 395 darts 241, 394
come up with 93 data 392
commander-in-chief 396 datum 392
commando 391 daughter 353
committee 245, 396 dawn 381
common 479 dear 476
528 Word index
deceased 485 dying 479,485

deep (adv) 490
deeply 490 each 157,180,243,246, 360-1, 363-
deer 240, 393 4, 378,390,400,429,441-4
delivery 386 each other 444
(the) departed 485 early 490,494
department 245 easily 489
deprive ... of 88 easy 153
derived 451 economic 449,451
deserve 214 economical 451
detached 479 edge 388
devil incarnate 351, 462 either 180,213,222-3,226,229,244,
devout 478 246, 249, 360-1, 363, 390,400-1,
diagnosis 393 436,443-4
dice 392 either... or 176-7,188,241-2
different 469,477 elder 471
difficult 153 eldest 477
dinner 381 elect 462
direct (adv) 490 elegantly 486
directly 490 elephant 398
dirty 457, 465 else 437-8
disappear 179 elsewhere 437
disappointed 219 embrace (n) 386
dishonest 452 end (n) 388,443
dissatisfied 449 English 449,458,465-6,471,485
do 39, 117, 204, 220-1, 235-8, 263, enough 375, 400, 436,462,448, 488
265,273,283,285,321 entail 248
dominant 451 entire 457
dominoes 394 entreat 254
double 182, 364 equal (v) 213
down 34, 86,124 erratum 392
downhill 491 estimation 386
downstairs 491 eternal 454
dozen 398 eternally 454
drake 354 even 226, 270
dreadful 470 even if 188,270
dress (v) 417 evening 381
drinkwise 490 events 380
drive 278 ever 140,213,228-9,294-5
drone (n) 354 evermore 491
drop 195 every 180,224,360-4,390,400,434,
drunken 451 436-7 439-40,434,442
dry (adj) 475 everybody 243,249,400,402,407,
duck 354, 398 436,439,441
due to 182 everyone 400, 436,439-42, 444
duly 489 everything 243, 400,435-6,441
Dutch 478,485 everywhere 436-7
dwarf 391 evidently 488
Word index 529
(the) Evil One 445 foremost 478

ewe 353 forever 303
except 64, 156,276,408,418 former 457-8,469,478
exert 417, the former 400,421
excluding (prep) 156 formula 392
expect 153 forthwith 491
extremely 448,456-7,487 fortunately 260
forward(s) 490
fact 248 four 469
faintly 487,489 fourth 364
fair (adj) 451 free (adj) 475
fairy 352 French 449,471,478,485
fall 53 friendless 451
family 245,396,420,430 friendly 451,488-9
far 476,490 functional 450
farther 476 funds 395
farthest 476 furniture 241
fast 478, 490 further 364, 476
father 353 furthermore 488
father-in-law 396 furthest 476
fears 395
feel 53,276 gallows 393
felt (participial) 479 gander 354
few 182,226-7,229, 357, 364, 390 gardens 395
400-1,436,469 gasworks 240
a few 226,401,445 general (adj) 469
fez 390 General Motors 358, 394
final 469 gentleman 353
find 153 gentleman thief 397
fine (adj) 475 genus 392
fire (v) 217 German 484
fireworks 395 get 53, 21,284-5
firm 245,420 girl 353
first 182, 364,400,421,465,469, give 179,204
477 give in 85
fiscal 471 give offence 95
fish (v) 278 give up 85, 187
five 363, 390 glasses 395
(a) flash of lightening 357 go on 264
flat (adj) 449,490 god 354
flatly 490 goddess 354
flock 366 goings-on 396
foodwise 490 golden 449
foot 392 good 438,449,457,465,476
for 175, 257,374 good-looking 452
for... sake 182,382 goods 395
forceps 395 goose 354, 392
foreign 450,458 government 245, 396,420,430
530 Word index
grab 205 heroine 354

great 478 hers 400,413
green 449,471 herself 400,402,416-18
greens 395 hesitancy 181
grey 475 hesitant 180
gross 393 hesitate 180
grotesque 478 high 457,478,480
grouse 393 high (adv) 490
grow 53 highly 407,448,487,490
(the) Gulf 372 him 400,404,407,414-5,417
himself 400,407,416-8
habeas corpus 11 hindmost 478
had better 282 his 361,400-1,407,413-5
the Hague 370 historic 452
half 182,363-4,400 historical 451
halfway 491 hit 195
halo 391 hold 153
hand-made 452 the Holy One 445
handsome 479 homewards 490
hang 124,278,454 hoof391
happen 43, 153 hopefully 488
happy 449, 479 horrible 470
hard (adj) 153,457,478 horse 353
hard (adv) 490 horsepower 393
hardly 140,226-7,229,490 hot 453
have 39,117,119,204,210,220-1, hourly 490
225,237,238,272,282-3,285-6, house 391
293,296,311 how 187, 190, 255, 372,400,424,
have got to 6 428,431-2,435
have to 6,284, 336 however 372
he 400-1,404,406-7, 408-11 huge 424
headquarters 240, 393 hundred 398
heap (v) 125 hundredweight 393
hear 215, 307 hypersensitive 452
heatedly 489 hypothesis 393
heavy 449-50,454 hypnotizable 471
heavily 454
The Hebrides 241,247,358, 394 I 272,399-400,404-6,408-10
heir apparent 351,462 identical 497
henceforth 491 if 10,119,183-4,187,189,255,270-1
hen-pheasant 355 if ...or 188,425
her 360-1,400-1,404,407,413,417, ill 476
420 imagine 276
herd 245 imaginable 460
here 124,286, 313,400,419,491 imbecile 484
hereabouts 491 (the) Imperial 371
hereinafter 4 imply 248
hero 354 in 64, 86,124,174, 187,491
Word index 531
in case 64,119,188,270 kill 179

in no way 260 kind of 351,375, 381
in order to 64 kindly 452
in order that 188 king 353
in spite of 182 know 153,195,215,271,298
in that 188
in view of 187 lady 352
incapable 449 larva 392
incompetent 452 last 364,446,476-7
index 393 late 476
(the) Indonese 485 later 476-7
indoors 491 latest 476
industrial 449, 471 (the) latter 400,421,476-8
(the) inevitable 403 lay 125
inferior 492 leading 471
influential 467 least 161,400,474,476,487
ingratiate 417 leg 358
(the) injured 484-5 less 161, 364,400-1,474,476-7,487
inmost 478 lesser 476-7
inner 478 lest 323
innermost 478 let 321
innings 393 liability 11
inside 491 lie 124, 300,454
insignificant 457 lien 11
instead of 187,266 light 449
interesting 467,479 like 264,432,418,478
internalized 471 linguistic 449
international 458,465 linguistics 394
(the) Irish 485 lion 354,398
is 406 lioness 354
it 44, 50-1,243-4,246,259,262-3, little 226-7,229,357,400,436,449,
268,400-1,404,406,410-4,421 472,474,476-7,490
its 246,360-1,400 a little 226, 364,401
itself 400, 416 livestock 241
the living 485
jackass 355 long 450,476,480,490
Japanese 393,485 look 53,276
jeans 241,395 look down on 93
jenny-ass 355 looker-on 396
Jew 354 looks 395
jokingly 489 lock (v) 201
junior 492 a lot of 351
jury 245 lots of 351
just 478 loud 489
loudly 489
keen on 448 lovely 470
keep 284-5, 300 low 457,478,490
Kew Gardens 248 lunch 381
532 Word index
Lyons 382 morning 381

most 161, 218, 364,400-1,448,473-
main 458 4,476,482,487
major 469 motto 391
majority 245 mouse 392
make 204 much 224,367,400,419,448,476,
make do with 95 487,491-2
man 354, 380, 392 much-debated 452
man-of-war 396 mumps 241, 394
man-servant 397 music 241
many 182,224, 357, 363-5, 390,400- must 6, 39, 283, 285, 325, 332-6,
2,419,436,444-5,469,476 338-9
mare 353 my 180,246, 361,400-1,413
marked 479 myself 400,415-6,418
marry 213
master 354 nanny-goat 355
mathematics 241, 394 narrow 457,475
matrix 393 nasty 470
matters 380 national 458
may 6, 39,119,220,232,269,283-5, native 457-8
290,316,323,325,327-8,332, navigable 455,459
334, 336, 339 near 491,476, 494
the Mayflower 371 nearer 476
me 360,400,404,408-9,414,417 nearest 476-7
mean 248, 298 nearby 491
means 240,393 nearly 493
measles 241,394 neat 475
media 395 need 6, 214, 284, 333-6
meet 168,213 negotiable 284
medical 449,457-8,465,471 neither 121,180,222- 4, 226,244,
meet 164,213 246,249,360-1, 363, 390,436,
mellow 451,479 443
mesmerized 479 neither ... nor 176-7,188,242
metal 471 nether 478
metropolis 393 the Netherlands 240
the Midlands 248 never 140,227,260,294-5
might 39,269-70, 328-9 nevertheless 260
million 398 Newcastle-on-Tyne 371
mine 400,402,413 news 241,344, 394
mistress 354 next 364,446, 469,476-7
monarch 353 nice 449,451
money 357 nicely 487,489
moneywise 490 night 381
monk 353 the Nile 371
monthly 489-90 no 180, 224, 360-1, 363,400-1,434,
moon 358 436^0,443
more 161,218, 357,364,400-1,436, no less 373
448,473-4,476,479,487 no one 224,249,400,436,439
Word index 533
noble 451 oneself 415,444

nobly 489 only 229,434,446,460,463,469,
noboby 224, 243, 260,400,436,438- 490
9 open (v) 195,201, 216
none 224,244,400,402,436,438, opposition 245
440-1,439-40,443,448 or 166,173-6, 178,189,241-2
none too 438 orders 395
nor 121, 176 the Orkneys 370
not 136,140, 177,219-22,224,227, ornate 478
237,486 other 477
not even 222-3,226 others) 364,400-1,436,443 A 446,
not only 177 449,465,469
nothing 224,260,400,436,438 ought to 6,283-4, 338-9
now 286, 313,400,494 our(s)400,413
nowhere 227, 260,437 ourselves 400,416
nuclear 471 out 86, 124
numerous 390, 469 outermost 478
nun 353 out of 187
outbid 278
oath 391 outdoors 491
oats 395 outright 491
oblong 449 outside 491
occasional 465 outskirts 395
occupy 300 over 86,491
octopus 392 oversleep 278
odds 395 owe 196
of 161, 387-9,413 A 430,438,440- own 198,413,415,445, 449
3,482-3,491 ox392
off 86
offhand 489 the Pacific 371-2
offhandedly 489 pain 395
offspring 393,491 pains 241,395
often 140, 494 pyjamas 241
old 450-1,472,476-7 palely 489
older 476-7 (a) pang of remorse 355
oldest 476-7 pants 395
on 86, 174,491 parents 353
on condition that 188, 270 party 395
once 184,364 passer-by 396
one 182, 224, 243,267, 357, 360-1, peace 241
363, 372, 377, 400-1, 403, 411, pence 392
419,441,445-6 people 394-6, 420
one(s) 436 per 378
ones 400 persuade 254
the one 423-4,436 phenomenon 393
the ones 423,436 phonetics 241
one another 444 pincers 395
one-fifth 182 plaice 393
534 Word index
pleasant 457,479 queen 353

please 432 quick 478
pleased 479 quiet 478
plenty of 351 quietly 489,494
pliers 395 quite 367, 373,448,487,493
pneumonia 379 quiz 390
Poet Laureate 351,396,462
poetic 451 ram 353
poetical 451 rapidly 489
police 241, 394-5 rarely 140,226-7, 229
polite 478 rather 122,218, 373,448,487,
politely 489 493
political 449,458,466,471 rather than 266
politics 394 read 195
the poor 389,403 real 478
popular 465-6 really 140,448,487
portmanteau 393 recently 294
the Portuguese 393,485 recognized 471
possess 210, 298 red 449,471
possibility 344 redback 4
possible 460 regrets 395
possibly 260 religious 449
post 381 remain 53
post-colonial 4 remains 395
Postmaster-General 396 remarkable 451
poultry 394-5 remarkably 487
pound 398 remember 264
poverty 241 remind 254
pragmatics 394 remind... of 88
pre-linguistic 452 republican 449
premises 395 request 254
present 460 require 214
president elect 351,462 resemble 213
presidential 457,471 resentment 241
presume 195 resistable 451
prettily 457,489 restless 451
pride 417 restrictive 451
primary 463,469,484 rice 357
problematically 489 ridicule 386
profound 478 the rich 389,401,483,485
promise(v) 100 riches 395
pronounced 479 rickets 394
proposal 344 right 478
prove 344 roast 216
provided (that) 188 round 449
publicly 489 royal 450
put 278 run 53
put up with 93 runner up 396
Word index 535
the Sahara 370,372 should 39,274,326, 334, 336-9

sake 388 shy 475
salmon 393 sibling 353
same 400,465,469,477 side 443
sands 395 significant 457
satellite 381 silent 454
satisfied 219,449 silently 454,489
savage 484 silk 471
save (prep) 156 silky 449
savings 395 silver 449
the Savoy 371 similar 469,477
say 211 similar to 448
scan 201 simple 153,479
scales 395 since 184,294
scarcely 226,229 sister 353
scarf 39 sit 124,300,454
schedule 211 skies 395
school 381 slacks 395
scissors 241 sleep 300
score 398 sleeping 471
scrub 4 slowly 487
second 182,363-4,400,469 sly 475
the second 421 small 457,478
secret 463 smallish 449,457,478
secretary general 35,462 smile 205
see 211,215, 307 snatch 195
seem 45, 53,153,218,276,452 snipe 393
seldom 226,229 so 121,176,222-3,226,372,400-1,
sentence (v) 100 448,459
senior (adj) 492 so ... as 474
sensible 449 so far 294
series 240,393 so that 188
several 182, 357, 364,400,444-5 Socrates 382
shall 6, 39,119,474,283,285,291, soft 457
325-6, 336-7 solar 449,457-8
shaken 479 solely 489
shambles 394 soluble 459
shatter 216 some 180,229,244,360-1,357,363,
shave 417 376,400-1,436-8,440-1,444
she 400, 404,406-10 somebody 229,243,260,400-1,407,
shears 241 436,438,440
sheath 391 somebody else 402
sheep 240, 353, 393,395 somehow 436-7
sheer 450,458 someone 229,400,407,436,440
(a) sheet of paper 357 something 229,243,400,436,438,
shingles 394 440
short 449,457 sometimes 229,437
show 217,278 someway 437
536 Word index
somewhere 229,260 supposing 119,270,322

son 353 sure 271
soon 490,494 surface 388
soothing 479 surroundings 395
sort of 351,375, 381 susceptible 471
sound 53,276 swap 397
species 393 swear 100
specific 469 sweep 205
spinster 353 (the) Swiss 393,478,485
spirit 395 switch 195
spirits 395 symposium 392
spokeswoman 354
spring 53 tabby-cat 355
spry 475 the Taj Mahal 371
square 449 take 204, 278
St. Paul's 385 take aback 211
staff 394-6,420 take after 88
stairs 395 take place 95, 182
stallion 353 tall 449-50,457
stand 124, 300,454 (a) tankard of beer 357
stand by 88 taste 53
start 195 the Täte 372
stateswoman 354 teach 52
statistics 394 team 396,420
statutory 11 tell 52,254
stay 53, 300 tend 43
still 176 the Thames 371-2
still life 391 than 161,408,418,479
stimulus 392 thanks 395
stone 357, 392-3 that 165,180,183-4,186,246,261,
stop 195, 264 360-1, 363, 390,412, 418-24, 428,
straight 490 430,432-6
straightforward 491 the 180,246, 360-1,400,402,415
(a) stroke of luck 357 their 360-1,400-1,413
strong 454,457 theirs 400,413
strongly 454 them 51,400,404,414,419
stupid 479 themselves 400,416-7
subsequent 469 then 270, 286,400,412
subtly 489 there 43-4, 124, 126, 244,286,400-1,
such 121,360-3,400-1,422 434,487 491
the Suez Canal 371 thereabouts 488,491
suggestion 344 thenceforth 491
(a) suit of armour 357 these 180, 360-1, 363, 390,400,419-
suitable 460 20,422,424
summer 381 they 400,402-5,407,410-2,418,420
superior 492 thick 478,480
supply ... with 88 things 380,454
suppose 153, 195,276 think 153, 276
Word index 537
third 182,364 unkind 479

this 180, 246, 313,360-1, 363, 390, unless 10, 119,270
400,412,418-24 unpleasant 457
those 180, 360-1, 363, 390,418-20, unthinkable 401
422-4,436 untidy 479
though 184,260,270 until 184
thousand 398 up 86,491
three 182,400,444 up to now 294
through 491 upon 491
thus 122 upper 478
tiger 354 uppermost 478
tights 241,395 upstairs 486
tigress 354 urge 254
till 184 us 400,404,409,414
the Times 371 used to 284
to 277,491 usual 458
today 294 utter 457
tom-cat 355
tomorrow 297 various 364
ton 398 vermin 394-5
too 222-3,226,229,373,448,487,490 vertebra 392
tooth 392 very 218,448-9,451,456-8,487,
topmost 478 490,492
touch 210 violently 489
tourist 471 visible 435,450,459
train 381 volcano 391
trousers 395
trout 393 wade 205
truly 489 wage(s) 395
truth 391 wait 300
turn 53 wait on 88,182
turndown 85 walled 452
twice 182 want 214-5,264
two 182, 400-1,444,469 was 239,406
two-thirds 182 watch 215
the Tyrol 370 waters 395
we 267,400,404-6,409-11,424
ugly 457 weak 457
the Ukraine 370 wear 300
unbind 278 weatherwise 490
uncle 353 weekly 490
under 491 well 122,476, 486, 490
underneath 491 well-to-do 13
under-sexed 452 wellnigh491
unhappy 449,452 were 119,239, 322,406
unicorn 352 (the) Welsh 485
the United States 248 westward(s) 490
the United Nations 248 wharf391
538 Word index
what 180,190,244,256,265,275, widow 354

360-3,400-1,423-8,430-1 widower 354
whatever 360-1,400,402,425,428, wild 463
431-2 will 39,220-1,274,285-6,290,292,
when 184,187,255,275,400,424, 305,311,319,325,339-40
428,431-2,435 winged 452
whence 424,428 winter 381
whenever 184,431 wish 322
where 184,187,400,424-5,428, with 134,174
431-2,435,491 woman 380, 392
whereas 184 woman doctor 397
whereby 428 wonderful 470
wherever 184,425,431 wooden 449,457,463
whereto 428 works 393
whereupon 184,275 worried 219,451
whether 183-4,187,189,255,424-5 worse 476
whether ... or 188 worst 476
which 180,186-7,244, 246,261, worth 338,448
274-5, 352, 360-1, 400-1,423-5, would 39,269, 274, 339-40
427-36 wreath 391
whichever 360-1,400,428,431-2 written 450
while 184 wrong 478
whilst 184
whither 424,428 Xerxes 382
who 186,244,261, 352,400-2,424-
30,432-6 yearly 489-90
whoever 400,425,428,431-2 yellow 449,471
wholly 489 yesterday 294
whom 186,400,424-6,428-9,432-5 yet 488,490,494
whose 180,186, 360-1,424-6,428, yoke 393
430 you 367, 399,404-5,410-1,444
whosoever 428 young 450-1,472,478
why 187,255,400,424-5,428,431- your 180,246, 361,400,413
2,435,491 yours 361,400, 413
wide 480 yourself 400,416,418,
widen 216 yourselves 416

(Topics in English Linguistics 22) Carl Bache, Niels Davidsen-Nielsen-Mastering English - An Advanced Grammar For Non-Native and Native Speakers-De Gruyter Mouton (1997) PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

(Topics in English Linguistics 22) Carl Bache, Niels Davidsen-Nielsen-Mastering English - An Advanced Grammar For Non-Native and Native Speakers-De Gruyter Mouton (1997) PDF

Uploaded by

Copyright:

Available Formats

Mastering English

© Printed on acid-free paper which falls within the guidelines of the

Library of Congress Calaloging-in-Publication Data

Bache, Carl, 1953-

Die Deutsche Bibliothek — Cataloging-in-Publication Data

© Copyright 1997 by Walter de Gruyter & Co., D-10785 Berlin

When in doubt about the acceptability of our examples we have conferred

Copenhagen and Odense, April 1997

Carl Bache & Niels Davidsen-Nielsen

Chapter 2: An introduction to syntax 13

Chapter 3: Elementary sentence analysis 29

3.2.6. The indirect object 50

Chapter 4: Advanced sentence analysis 71

5.2.3. Information structure 113

Chapter 6: Coordination and subordination 165

6.2.1. Coordination and ellipsis 167

Chapter 7: The simple sentence 191

7.5.8. Semantic scope of negation 230

Chapter 8: The complex sentence 253

9.6.9. Perfect progressive forms 305

Chapter 10: Nominals 341

10.4. Quantification: the number category 390

Chapter 11: Pronominale 399

Chapter 12: Adjectivals and adverbals 447

12.3.1. The basic system of comparison 473

Subject index 503

Word index 525

1.1. Goals and framework

constituents, complex predicators and the relationship between sentence type

written texts (newspapers, magazines, fiction, etc.), from other grammars of

1.4. Varieties of English

1.5. Standard English

1.6. Grammatical variation

1.7. Variation according to medium

speech, catching the receiver's attention cannot be done prosodically or

1.8. English for Special Purposes

2.1. The word

2.2. The sentence

(6) I say no!

2.3. Grammatical structure

2.4. Linearity and the principle of proximity

(7c) Stephanie, Alex and Roger went sailing this morning.

group (according to the principle of proximity) and as such enter a 'joint'

2.6. Linguistic creativity and ambiguity

(1) Old men and women are invited to the party.

units making up the sentence. The recognition of ambiguity in such cases is

2.7. Competence and performance

When describing a language we are therefore concerned not only with

2.8. Syntagmatic and paradigmatic relations

The choice of verb form in a frame like [Sally grammar], where a

3.1. The basic form and function approach

3.1.1. Form and function

3.1.2. Sentence functions

The tree diagram, which is a conventional form of syntactic representation,

The lines slanting downwards from Sent indicate a 'consist-of relationship1:

The old man wrote a long letter

(2) He was writing very slowly.

(3) The letter was unbearably long.

The letter was unbearably long

3.1.3. Four form types

That Helen left had bothered Jack and Jill immensely

The function-over-form convention is used to indicate that a sentence

form type to be further analysed 'downwards in the tree' (unless of course it